9

Kinetic theory
9-1 INTRODUCTION

9-2 BASIC ASSUMPTIO NS

9-3 MOLECULAR FLUX

9-4 EQUATION OF STATE OF AN I DEAL GAS

9-5 COLLISIONS WITH A MOVING WALL

- THE PRINCIPLE OF EOUIPARTITION OF ENERGY

9-7 CLASSICAL THEORY OF SPECIFIC HEAT CAPACITY

9-8 SPECIFIC HEAT CAPACITY OF A SOLID

 250 KINETIC THEORY

8-1 INTRODUCTION
The subjeJt of thermodynamics deals with the conclusions that can be drawn from
certain experi mental laws, and with the applications of these conclusions to rela-
tions between properties of materials such as heat capacities, coefficients of ex·
pansion, compressibilities, and so on. It makes no hypotheses about the nature of
matter and is purely an empirical science.
Although thermodynamic principles can predict many relations between the
properties of matter, such as the difference between the specific heat capacities
cp and c., or the variation of these quantities with pressure, it is not possible to
derive from thermodynamic considerations alo ne the absolute magnitudes of the
heat capacities, or the equation of state of a substance.
We can go beyond the limitation of pure thermodynamics only by making
hypotheses regarding the nature of matter, and by far the most fruitful of such
hypotheses, as well as one of the oldest, is tha t matter is not continuous in structure
but is composed of particles called molecules. I n particular, the molecular theory
of gases has been very completely developed, because the problems to be solved
are much simpler than those encountered in dealing with liquids and solids.
The properties of matter in bulk are predicted, starting with a molecular theory
by means of two different lines of attack. The first, called the kinetic or dynamic
theory, applies the laws of mechanics to the individual molecules of a system, and
from these laws derives, for example, expressions for the pressure of a gas, its
internal energy a nd its specific heat capacity. The method of statistical thermo-
dynamics ignores detailed considerations of molecules as individuals, and applies
considerations of probability to the very large number of molecules that make up
any piece of matte r. We shall see that the methods of statistical thermodynamics
provide a further insight into the concept of entropy and the principle of increase
of e ntropy.
Both kinetic theory and statistical thermodynamics were first developed on the
assumption that the laws of mechanics, deduced from the behavior of matter in
bulk, could be applied without change to particles like molecules and electrons.
As the sciences progressed, it became evident that in some respects this assumption
was not correct; that is, conclusions drawn from it by logical methods were not in
accord with experimental facts. The failure of small-scale systems to obey the
same laws as large-scale systems led to the developmen t of quan tum theory and
quantum mechanics, and statistical thermodynamics is best treated today from the
viewpoint of quantum mechanics.
This chapter and the next will be devoted to the kinetic aspects of molecular
theory, and the following chapters to statistical thermodynamics. As we go along,
we shall make many references to concepts and equations that have already been
discussed in the preceding chapters on thermodynamics, and we shall see how a
much deeper insight into many questions can be attained with the help of molecular
theory as a background.
 &-2 BASIC ASSUMPTIONS 251

8-2 BASIC ASSUMPTIONS

In thermodynamics, the equation ofstate of a system expresses the relation between
its measu rable macroscopic properties. The simplest equation of state is that of an
ideal gas; and although kinetic theory is limited neither in concept nor in applica-
tion to ideal gases, we shall begin by showing how the equation of state of an ideal
gas can be derived on the basis of a molecular model with the following as-
sumplions:
I. Any macroscopic volume of a gas contains a very large number of molecules.
This assumption is justified by all experimental evidence. The number of molecules
in a kilomole (Avogadro's• number N..) is 6.03 x 10". Experimental methods
for arriving at this number are discussed in later chapters. At standard conditions,
I kilomole of a n ideal gas occupies 22.4 m1. Hence at standard conditions there are
approximately 3 x 10" molecules in a cubic meter, 3 x 1011 in a cubic centimeter,
and 3 x 1011 in a cubic millimeter.
2. The molecules are separated by distances that a re large compared with their own
dimensions and are in a state of continuous motion. The diameter of a molecule,
considered as a sphere, is about 2 or 3 X IQ-10 m. If we imagine one molal volume
at standard conditions to be divided into cubica l cells with one molecule per cell,
the volume of each cell is 30 x JQ-17 m•. The length of one side of suchla cell is
about 3 X I Q-1 m, which means that the distance between molecules is of this o rder
of magnitude, about 10 times the molecular diameter.
3. To a first approximation, we assume that molecules exert no forces on one
another except when t hey collide. Therefore between collisions wi th other mole-
cules or with the walls of the container, and in the absence of external fo rces, they
move in s traight lines.
4. Collisions of molecules with one another and with the walls are perfectly elastic.
The walls of a container can be considered perfectly smooth, so that there is no
change in tangential velocity in a co llision with the walls.
S. In the absence of external forces , the mo lecules are distributed uniformly,
throughout the container. If N represents the total numbe r of molecules io a
container of volume V, t he average number of molecules per un it vol ume, n, is

n - N/V.
The assu mption of uniform distributio n then implies that in any ele ment of
volume AV, wherever located, the number of molecules AN is

AN= nAV.
Obvio usly, the equation above its not correct if AVis too small, since the number
of molecules N, although large, is finite, and one can certainly imagine a volume

• Count Amedeo Avogadro, Italian physicist (1776-1856).

 2$2 KINETIC THEORY 9-2

element so small that it contains no molecules, in contradiction to the equation

above. However, it is possible to d ivide a container into volume clements large
enough so that the number of molecules per unit volume within them does not
difl'cr appreciably from the average, and at the same time small eno ugh compared
with the di mensions of physical apparatus that they can be treated as infinitesimal
in t he mathematical sense and the methods of difl'crcntia l and integral calculus
can be applied to them. For example, a cube 1/ 1000 monon a side is certainly
small in comparison with the volume of most laboratory apparatus, yet at standard
conditions it contains approximately 30 million molecules.
6. The directions of molecular velocities a re assumed to be distributed uniformly.
To put t his assumption in analytic form, imagine tha t there is attached to each
molecule a vector representing the magnitude and direction of its velocity. Let us
t ransfer all these vectors to a common origin and construct a sphere of a rbitrary
radius r with center at the origin. The velocity vectors, prolonged if necessary,
intersect the surface of the sphere in as many points u there are molecules and the
assumption of uniform distribution in direction means that these points are distrib-
uted uniformly over the surface of the sphere. T he average number of these
points per unit area is

and the number in any element of area AA is

N
AN- ,,• AA,
4

wherever the e lement is located. As in the preceding paragraph, the area must be
large enough (that is, it must include a large enough range of directions) so that
the surface density of points within it does not difl'er appreciably from the average.
Because of ~he large number of molecules, the range of di rections can be made
very small and still include a large number of points.
Let us carry this description of velocity directions o ne step further. Any
arbitrary direction in space can be specified with reference :o a polar coordinate
system by the angles 0 and ,P, as in Fig. 9-1. The a rea AA of a small element on
the surface of a sphere of radius r is, very nearly,

AA = (r sin 0 AO)(r A,P) = r• sin 0 AOA,P.

The nu mber of points in this area, or the number of molecules AN,, having
velocities in a direction between 0 and 0 + AO, .p and .p + A,P, is
 9-2 BASIC ASSUMPTIONS 253

When bolh sides of this equation are divided by the volume V occupied by the gas,
we get
An,~ = !!.. sin 0 AOAt/>, (9-1)
411
where A.n,• is the number density of molecules with velocities having directions
between 0 and 0.+ AO, and 4> and 4> + At/>.

rJjn8 rsinOA'

Fig. 9- 1 Polar coordinates.

Consider, finally, the magnitudes of the molecular velocities, or the spuds

of the molecules. It is clear that not all molecules have the same speed, although
this simplifying assumption is onen made. Even if we could start them off in this
way, intermolecular collisions would very quickly bring about differences in speed.
We shall show in Seclion I 2- 2 how to calculate the number that have speeds in
any specified range, but for the present we shall assume that the speed can have
any magnitude from zero to infinity• , and we represent by AN. the number of
molecules with speeds between v and v + Av. Geometrically this number equals
the number of velocity vectors tuminating within a thin spherical shell in Fig. 9-1,
between spheres of radii r 1 . . o and ' • • v + Av. As a result of collisions, the
speed of any one molecule is continually changing, but we assume that in the equi-
librium state the number of molecules with speeds in any specified range remains
constant.

• It would be beuer to say, from zero lo the speed oflighl. However, as we shall show, 1he
number of molecules with speeds exceeding even a small fraction of the speed of light is so
small for ordinary gases that for mathematical simplicity we may as well make the
assumption above.
 254 KINETIC THEORY

$-3 MOLECULAR FLUX

Because of the continuous random motion of the molecules of a gas, molecules are
continually arriving at every portion of the walls of the container, and also at each
side of any imagined surface within the gas. Let I:!.N represent the total number of
molecules arri ving from all directions and with all speeds at one side of an element
of surface of a rea I:!.A during a time interval l:!.t. The molecular flux Cl> at the surface
is defined as the total number of molecules arriving at the surface, per unit area and
per unit time. Thus,
(9-2)

lf the surface is an imagined one within the gas, all molecules arriving at the
surface, from either side, will cross it, and if there is no net rr.otion of the gas as a
whole, the molecular fluxes on either side oft he surface are equal and are in opposite
directions. Thus at either side of the surface there are two molecular fluxes, one
consisting of molecules arriving at that side and the other consisting of molecules
that have crossed the surface from the other side.
lf the s urface is at the wall of the container, molecules arriving at the surface
do not cross it but rebound from it Hence there are also two fluxes at such a
surface, one consisting of molecules arriving at the surface and the other consisting
of molecules rebounding from the surface.
In Fig. 9-2, the shaded area I:!.A represents a small element of surface, either
within the gas or at a wall. Construct the normal to the area, and some reference
plane containing the normal. We fi rst ask, how many molecules arri ve at the
surface during a time interval l:!.t, travelling in the particular direction 8, <f>, and
with a specified speed v. (To avoid continued repetition, Jet it be understood that
this means the number of molecules with directions between 8 and 8 + 1:!.8, </>and
</> + 1:!.¢>, and with speeds between v and v + l:!.v.)
Construct the slant cylinder shown in Fig. 9-2, with axis in the direction 8,
¢>, and of length v l:!.t, equal to the distance covered by a molecule with speed v
in time l:!.t. Then the number of 0</>o-molecules that arrive at the surface during
the time l:!.t is equal to the number of O<f>v-molecules in the cylinder, where a 8</>o-
molecule means one with speed v, traveling in the 0, </>direction.
T o show that this is correct, we can see fi rst that all 0</>v-molecules in the cy-
linder will reach the surface during the time l:!.t. (We are ignoring any collisions
with o ther molecules that may be made on the way to the surface, so tha t the
molecules are considered as geometrical points. In Section 10-3 we shall see how
to take such collisions into account.) There are, of course, many other types of
molecules in the cylinder. Some of these will reach the surface element during the
time 1:!.1 and others will not. Those that do not are either not traveling toward the
element (that is, they are not O<f>-molecules) or are not traveling fast enough to
reach the element during the time l:!.t (that is, their speed is less than v). Those
within the cylinder that do reach the surface during the time l:!.t are necessarily
8¢>-molecules, but unless they have a speed v they are not 8¢>o-molecules.
 MOLECUlAR FLUX 255

Fig. 9-1 Only the Btlv-molecules in the cylinder will arrive at the
an:a ll.A during a time ll.t.

Many ollter molecules, not in the cylinder, will arrive at the element during
the time I:J.t. Some of these will have a speed u, but they are not 8<f>-molecules since
they come in from other directions. Therefore al/8~v-molecules in Lite c:ylinder,
and only those molecules, will reach the surface during the time I:J.t, traveling in the
0</>-direction with speed u.
Let 6.n, represent the number density of molecules with speeds between v and
v + tJ.u. Then from Eq. (9- 1) the number density of O~molecules is

lin,., - ..!.. lin, sin 0 68 li~. (9-3)

4,.
The volume of the slant cylinder in Fig. 9-2 is
liV = (liA cos O)(v lit).
The number of 8~v-molecules in the cylinder is therefore

· liN_.,=..!.. ulin, sin 8 cos 8 MJ li~liA lit,

4,.
and the ftux li4>1#• of O~v-molecules is

li4>_., - liN,., oa _!_ vlin,sin Ocos OliO li~. (9-4)

liAlit 4,.
 268 KINETIC THEORY

The flux t.<Il,. due to molecules arriving at an angle 0 with speed v, but in-
cluding all angles tf>, is found by replacing t'.tf> with dtf> and integrating over all
values of tf> from 0 to 27T. The result is

t.<I~,. =- ~ v t.n. sin 6 cos 6 t.6. (9-S)

The flux t.<Il, due to molecules arriving at the angle 6, including all angles
tf> and all speeds v, is found by summing the expression for t.<%1,. over all values of
v. Thus
t.<%11 • ~ sin 6 cos 6 t.6 2 v t.n., (9- 6)

The flux t.<Il. of molecules with speed o, including all angles 6 and tf>, is found
by replacing t.6 with d8 in Eq. (9-5) and integrating over all values of 6 from zero
to "/2. This gives
I
t.<%1. = D t.n., (9-7)
4
Finally, the total flux <II, including all speeds and all angles, is obtained either
by summing t.<ll. over all values of v, or by replacing t.O with dO in Eq. (9-6) and
integrating over 8 from zero to 7T/2. The result is

<ll - !4 2 v t'.n., (9-8)

Let us express this result in terms of the avuagt or aritlzmttic mean speed 0.
This quanfity is found by adding toget~er the speeds of all the molecules, and
dividing by the total number of molecules:
- 2v
v=N,
where the sum is over all moltcults. But if there are t'.N1 molecules with speeds v,.
t'.N, molecules with speeds v,, etc., the sum of the speeds can also be found by
multiplying the speed v1 by the number of molecules t'.N1 having that speed,
multiplying v, by the number t'.N, having speed v,, and so on, and adding these
p roducts. The average speed is then the sum of all such products, d ivided by the
total number of molecules. That is,
v=o,t'.N,+v,t'.N,+···=..!.~vt'.N (9-9)
N N"- ••
where the sum is now over all spttds. When nume rator and denominator are
divided by the volume V, we have

It follows that
2vt'.n,- un, (9-10)
 MOLECULAR j LUX 1&7

and hence from Eq. (9-8) the molecular flux ~. including all molecules arriving
at one side of the element and coming in from all directions and with all speeds, is

I.
~ ~ vn. (9-1 1)
4
As a numerical example, the number of molecules per cubic meter, n, is approxi-
mately 3 x IOU molecules m-.1 at standard conditions. We shall show later that the
average speed of an oxygen IT!olecule at 300 K Is approximately 4SO m s-1• The
molecular flux in oxygen at standard conditions is therefore
I .J
4> • 4no "' 4 X 3 X JQU X 450 "' 3.3 X 1017 molecules m- 1 s-•.

It is sometimes useful to put Eq. (9-4) in the following form. Consider the
area t.A in Fig. 9-2 to be located at the origin in Fig. 9-1 and to lie in the x-y
plane. The molecules arriving at the area in the 8,P-direction are those coming in
within the small cone in Fig. 9- 1, whose base is the shaded area t.A on the spherical
surface in that diagram. This area is
t.A "" r' sin 8 M t..p,
and the solid angle of the cone, !J.w is

!J.w = !J.: = sin 8 !J.8 !J.,P.

r
Hence from Eq. (9-4) the flux !J.011, can be wrillen

!J.0 1 f , = ..!... vAn, cos 8 Aw = t.0.,; (9- 12)

47T

and the flux per unit solid angle, of molecules with speed v, is

A4>~, = ..!. vAn, cos 0. (9-13)

t.w 4>r

!he total flux per unit solid angle, including all speeds, is

M>M- ...!.. lin cos 0. (9-14)

!J.w 4>r

If we consider a number of small cones with apexes at AA in Fig. 9-1, the

greatest number of molecules arrives with direction in the cone centered about the
normal, since cos 0 has its maximum value for this cone, and the number decreases
to zero for cones tangent to AA, where 0 .. 90°.
258 KINETIC THEORY

If the area D.A is a hole in the wall of a thin-walled container, small enough so
that leakage through the hole does not appreciably affect the equilibrium of the
gas, then every molecule coming up to the hole will escape through it. The distri-
bution of directions of the molecules emerging from the hole is also given by Eq.
(9- 14). The number emerging per unit solid angle is a maximum in the direction
normal to the plane of the hole and decreases to zero in the tangential direction.

1-4 EQUATION OF STATE OF AN IDEAL ~AS

Figure 9-3 shows a 8,Pu-molecule before and after a collision with the wall of a
vessel containing a gas. From our assum ption of perfect elasticity, the magnitude
of the velocity v is the same before and after the collision, and from the assumption
that the wall is perfectly smooth, the tangential component of velocity is also un-
altered by the collision. It follows that the angle of reflection is equal to the angle
of incidence and the normal component of velocity is reversed in the collision,
from v cos 8 to - v cos 8.

Fig. 9-3 Change in velocity in an elastic collision.

 EQUATION OF STATE OF AN IDEAL GAS 259

The force exerted on the wall by any one molecule in a collision is an impulsive
force of short duration. The details of its variation with time are unknown; but it
is not necessary to know them because from Newton's second law we can set the
average force per unit area exerted on the surface, or the average pressure, equal to
the average rate of change of momentum per unit a rea.
If m is the mass of a colliding molecule, the change in the normal component
of momentum in a 8,Pv-collision is

mv cos 8- (-mv cos 0) = 2mv cos 8. (9-15)

The change in momentum depends on 8 and v, but not on the angle tf>. Hence
we need the number of 0<>-molecules arriving at the surface per uni t area, a nd
per unit time, or the flux t.<ll,. given by Eq. (9- 5).
The rate of change of momentum per unit area due to all molecules arriving
at an angle 0 with speed v, or the pressure t.P,., equals the product of t.4>,. and
the change in momentum of a 8v-molecule:

t.P,, = (iv t.n. sin 8 cos 8 t.8)(2mv cos 8) = mv' t.n. sin 8 cos' 8 t.O.
To find the pressure t.P. due to molecules of speed v coming in at all values of 0,
we integrate over 0 from 0 to "'/2. This gives

t.P. = 3I mv' t.n,.

Finally, summing over a ll values of v, we have for the total pressure P,

(9- 16)

The same reasoning as that above can be applied to any imagined surface in
the inte ri or of the gas. The molecular flux t.<ll,.., is the same for a ll surfaces,
wherever located. Molecules approaching an internal s urface from one sjde pass
through it and do not rebound, but the flux across the surface from the opposite
side carries the same momentum away from the surface as do the molecules re-
bounding from a wall of the container. That is, for every 8</>v-molecule crossing
the surface from o ne side, there will be another O,Pv-molecule crossing from the
other side, and Fig. 9-3 will apply to any surface wi thin the gas, except that the
black circles in Fig. 9- 3 do not represent the same molecule.
Hence the net flux of momentum , at right angles to any surface, is the same as
at the boundary wall; and if we consider the pressu re as the flux of momentum,
the pressure has the same value at all points, both within the gas and at its surface.
Equation (9- 16) is more conveniently expressed as follows. Tbe average value
of the square of the speed of all molecules, or the mean square speed, is found by
 260 KINETIC THEORY 9-4

squaring all the speeds, adding these quantities, and d ividing by the total number of
molecules:

Just as in calculating the average speed, we can obtain l:v1 more conveniently
v: v:
by multiplying by I!.N., by 6N1 , etc., and adding these products. That is
I
01 =I v1 1!.N. or
N
Then

and
I -
P-
3nmv'. (9-17)

Since the mean kinetic energy of a single molecule is •mV', the right side of Eq.
(9- 17) equajs two-thirds of the total kinetic energy per unit volume or two-thirds
of the kinmc ~ntrgy dmsity; and Eq. (9-17) thus expresses the pressure in te rms of
the kinetic energy density.
It will be shown in Section 12- 2 that the average value of the square of the
speed, v', is always greater than the square of the average speed, (<i)1•
Since n represents the number of molecules per unit volume, N/ V, we can
write the p receding equation as
I -
P V - JNmv'.

This begins to look like the equation of state of an ideal gas,

PV = nRT,

where n represents the number of kilomoles , equal to the total number of mole-
cules divided by the number of molecules per kilomole, or Avogadro"s number N...
We can therefore write the equation of state of an ideal gas as

PV- N~ T.
N ..
The quotient R/N.. occurs frequently in kinetic theory. It is called the universal
gas constant per molecule, or Boltzmann's constam, a nd is represented by k:

k=~. (9- 18)

N._
 EQUATION OF STATE OF AN IDEAL GAS 261

Since Rand N._ a re universal constants, k is a universal constant a lso. T hat is,
its magnitude depends only on the system of units employed. In the MKS system,

k = ~= 8 ·314 x IO' = 1.381 x 10_,. J molecule-• K-l.

N4 6.022 X IO"
In terms of the Boltunann constant, the equatio n of state of an ideal gas
becomes
PV - NkT.
T his will agree with the equa!ion derived from kinetic theory, Eq. (9-17), if we set

I -
NkT =- Nmv',
3
or
;; - 3kT. (9-19)
m
The theory has thus led us to a goal we did no t deliberately set out to seek;
namely, it has given us a molecular interpretatio n of the concept of absolute
temperature T, as a quan tity proportional to the mean square speed of the
molecules of an ideal gas. It is even more s ignificant to write Eq. (9-19) as

-I mv'
- 3
=- kT. (9- 20)
2 2
The product of one-half the mass of a molecule and the mean square speed is the
same as the mean translational kinetic energy, and we see fro m the preceding
equation that the mean translational kinetic energy of a gas molecule is pro-
portional to the absolute temperature. Furthermore, since the factor 3kf2 is t he
same for a ll molecules, the mea n kinetic energy depends only on the temperatu re
and not o n the pressure o r volume o r species of molecule. That is, t he mean kinetic
energies of the molecules of H,, He, 0 ,, Hg, etc., are all the same at the same tem-
perature, despite the disparities in their masses.
We can compute fro m Eq. (9-20) what this energy is at any tempera ture. Le t
T - 300 K. Then
3 3
2kT= i X 1.38 X 10- u X 300 = 6.21 x IQ-"J.

If the molecules are oxygen, the mass m is 5.3 1 x IO-" kg, a nd the mean
square speed is
.... 2 x 6.21 x ro-" , , ,
.,.. • . x _ii - 23.4 x 10 m s- .
5 31 10
The square root of this quantity, or the root·m•an·squau <peed is

•,... - ~- 482 m s-1 • 1607 ft r 1

- I IOO mi hr-1•
 282 KINETIC THEORY &-5

By way or comparison, che speed or sound in air al slandard condicions is about

350m s-1 or 1100 rc s-1 and lhe speed or a .30 cal rifle bulle! is aboul 2700 rc s-1•
The speed of a compressional wave in a fluid is given by

v =~
which, for an ideal gas, is equivalent to

v- .JykT/m,
where y = cpfc•. Since the root-mea n-square speed of a molecule is

v,.,. = .J3kT/m, {9-21)

we see t hat the two a re nearly equal but that the speed of a sound wave is somewhat
smaller than the rms molecular speed, as would be expected.
When electrons and ions are accelerated by an e lectric field , it is convenient to
express their energies in tlectron-IJ{)/ts {abbreviated eV), where by definition

l electron-volt • 1.602 x 10""11 J.

An e lectron-volt is equal to the energy acquired by a particle of charge e =

1.602 x 10""11 C accelerated through a potential difference of I V.
At a te mpe rature of 300 K,

l3 kT • 6.21 X 10""11 J ~ 0.04 eV.

or
kT = 0.026 eV ~ -h eV.
Hence at a temperature of 300 K, the mean kinetic e nergy of a gas molecule is
only a few hundredths of an electron-volt.

9- 5 COLLISIONS WITH A MOVING WALL

We now examine the nature of the mechanism by whic h an expa nding gas does
work against a moving piston, a nd show that if the process is adiabatic, the work
is d one at the expense of the kinelic energy of the molecules (that is, the internal
energy of the gas) and the temperature of the gas decreases. Figure 9-4 represents
a gas in a cylinder provided with a piston. Let the piston move up with speed u,
small compared with molecular speeds and small enough so that the gas remains
practically in an equilibrium state. From the thermodynamic viewpoint, then, the
process is reversible.
When a molecule collides elastically with a stationary wall, the magnitude of
the normal component of velocity is unchanged. If the wall is moving, the mag-
nitude of the relative velocity is unchanged.
 9-5 COLLISIONS WITH A MOVING WALL 283

. ;; '
;/,· ~~

Fig. 9-4 Collisions with

a moving wall.

To illustrate by a simple numerical example, if a particle approaches a stationary

wall normally with a speed of ISm s-•, referred to a coordinate system fixed in the
laboratory, it rebounds with a speed of ISm s-•. If the wall is moving away from the
particle with a speed of 5 m s-•, and if the particle has a speed of 20 m s-•, both
relative to the laboratory coordinate system, the molecule is again approaching the
wall with a relative velocity of ISm s-•. After the collision the magnitude of the
velocity of the particle relative to the wall will again be IS m s-1 , but since the particle
is now moving in a direction opposite to that of the wall, its speed in the laboratory
coordinate system is only 10m s- •.
In general, if t he normal component of the velocity before collision is v cos 8,
where 8 is the angle between v and the normal to the wall, the velocity component
after collision, v' cos 8' , is equal to v cos 6 - 2u. The loss of ki netic energy in the
collision is
I I
m(v cos 6)1 - m(v cos 6 - 2u)1 ~ 2mvu cos 8,
2 2
since by hypothesis u « v. The kinetic energy of the molecule can decrease even
if the collision is perfectly elastic, because in the collision the molecule exerts a
force against a moving wall and he nce does work on the wall.
The loss of kinetic energy depends on 6 a nd v but not o n ~· By Eq. (9-S) the
number of Ov-collisions with a wall, per unit area and per unit time, is

A<ll,. - ~ v 6n, sin 8 cos 6 60.

Multiplying this by the loss in kinetic energy in s uch a collision, we obtain for the
loss in kinetic energy per unit area and per unit time, by molecules making 8v-
collisions,
muv• 6n, sin 6 cos• 6 60.
 264 KINfTIC THfORY 9-6

Finally, after integrating over 0 from 0 to "/2, and summing over all values of v,
we get

for the total loss of molecular kinetic energy, per unit area and per unit time. But
inm'iii equals the pressure P, and if the area of the moving piston is A, the decrease
of molecular kinetic energy per unit time is

PAu = Fu . (9-22)

The product Fu (force times velocity) gives the rate at which mechanical wo rk
is d one on the piston or the power developed by the expanding gas, and we see that
this is j ust equal to the rate of decrease of molecular kine tic energy. If the molecules
do not reCfive ene rgy from any other source, their kinetic energy, and hence the
temperature of the gas, decreases. Note that it is not correct to say that the tem-
perature of a moltcule decreases. From the molecular point of view, temperature
is a n attribute of the assembly of molecules as a whole, namely, a quantity pro-
portional to the mean kinetic energy. An individual molecule can have more or
less kinetic energy but it does not have a higher or lower temperature.
The derivation above was based on the assumption that the piston velocity, u,
was very much smaller than the molecular velocities, and it does not hold if the
piston is pulled up rapidly. In particular, if the piston velocity is very much greater
tha n the molecular velocities, no molecules (or at least very few) will be able to
overtake the piston and collide with it. Then there is no loss of kinetic energy and
no decrease in temperature, intermolecular forces being neglected. Such a process
is equivalent to an expansion into a vacuum, as in the Joule experiment, where we
showed on thermodynamic grounds that the work and the change in internal
energy were both zero.

11-1 THE PRINCIPLE OF EOUIPARTITION OF ENERGY

Suppose we have a mixture of gases that do not react chemically with one another,
and that the temperature and density are such that their behavior approximates
that of an ideal gas. It is found experimentally that the total pressure of the
mixture is the sum of the pressures that each gas alone would exert if a mass of
each, equal to the mass o f that gas in the mixture, occupied the entire volume of
the mixture. The pressure that would be exerted by each gas if present alone is
called its pattial prtssurt and the experimental law above is Dalton's law ofpartial
pussuru. If the gases are distinguished by subscripts, we can then write
p, V - N 1k T, etc.,
where p 1 , p,, etc. are the partial pressures of the constituent gases, N1 , N,, etc.
 THE PRINCIPLE OF EOUIPARnTION OF ENERGY 285

are the numbers of molecules of each constituent, and V and Tare the volume and
temperature, common to all of the gases.
Let m., m,, etc. represent the masses of the molecules of the constituents
and v~. v:O etc., the respective mean square speeds. By the methods of Section 9-4,
considering the collisions of each type of molecule with the walls and computing
the pressure produced by each, we would find

p,V • 3I N,m,v~.
-
p1 V - 3I N,m,v~,
-
etc.

Equating corresponding expressions for p. V, p,V, etc., gives

2I m,v~- = 32kT, ! m,;;: - ~ kT, etc.

2 2
The terms on the left side of the preceding equation are the mean translational
kinetic energies of the molecules of the various gases, and we conclude that in a
mixture of gases the mean kinetic energies of the molecules of each gas are the same.
That is, in a mixture of hydrogen and mercury vapor, although the masses of the
molecules are in the ratio of 2 to 200, the mean translational kinetic energy of the
hydrogen molecules equals that of the mercury molecules.
The example above is one illustration of the principl~ of ~quiportitlon of en~rgy.
We know now that this principle is not a universal law of nature but, rather, a
limiting case under certain special conditions. However, it has been a very fruitful
principle in the development of molecular theories.
Let us give another example. The translational kinetic energy associated with
the x-component of the velocity of a molecule of mass m is lmv!, with corre-
sponding expressions for they- and z-components. The mean square value of the
velocities of a group of molecules is

;.=;:+;;:+;;:.
Since the x-, y-, and z-directions are all equivalent, the mean square values of the
components of velocity must be equal, so that

and
;. = 3v! = 3;: -= 3~.
The mean kinetic energy per molecule, associated with any one component of
velocity, say v., is therefore
!2 m;.• - !6 m;. - !2 kT.
Since the mean total translational kinetic energy per molecule is 3kT/2, it follows
that the translational kinetic energy associated with each component of velocity
is just one-third of the total.
 266 KINETIC THEORY 9-8

Each independent quantity that must be specified to determine the energy of a

molecule is called a degree offreedom. Since the translational kinetic energy of a
molecule is determined by the three velocity components of its center of mass,
it has three translational degrees of freedom. We see that the ave rage translational
kine tic energy per molecule is divided equally amorig them. In other words, we have
equipartition of the energy among the three translational degrees of freedom.
. Molecules, however, are not geometrical points but are o f finite size. They
have moments of inertia, as well as mass, and can therefore have kinetic energy
of rotation as well as of translation. Furthermore, we would expect them to rotate
because of the random collisions with other molecules and with the walls. Since
the angular velocity vector of a rotating molecule can have a component along all
three coordinate axes, a molecule would be expected to have three rotational
degrees of freedom or, if it is a rigid body, six degrees of freedom in all. However,
molecules are not perfectly rigid structures and can also be expected to oscillate
or vibrate as the result of collisions wi th other molecules, giving rise to still more
degrees of freedom. (It may be mentioned at this point that rotations and vibrations
of molecules are facts that are as well established as most of our other information
about molecular properties. The best experimental method of studying rotations
and vibrations consists of a spectroscopic analysis of the light emitted or absorbed
by molecules in the infrared.) Without committing ourselves to any specific number,
let us say that in general a molecule hasf degrees of freedom, of which 3 only are
translational, however complex the molecule.
We shall show in Section 12-5, on the basis of the Boltzman n statistics, that if
the energy associated with any degree of freedom is a quadratic function of the
variable specifying the degree of freedom, the mean value of the corresponding
energy equals kT/2. For example, the kinetic energy associated with the velocity
component v. is a quadratic function of v., and, as shown above, its mean value
equals kT/2. Similarly fo r rotat i~n. whe re the kinetic energy is Jw'/2, the mean
rotational kinetic energy is kT/2; and for a harmonic oscillator, where the potential
energy is Kx'/2 (K being the force constant), the mean potential energy is kT/2.
Hence all of the degrees of freedom for which the energy is a quadratic fu nction
have associated with them, on the average, equal amounts of energy; and if all
degrees of freedom are of this nature, the total energy is shared equally among
t hem. This is the general statement of the principle of equipartition of energy.
The mean total energy of a molecule withfdegrees of freedom, assuming the equi-
partition principle holds, is therefore

i - £2 kT, (9-23)
and the total energy of N molecules is

Ni = £ NkT = [_ nRT, (9-24)

2 2
where n is the number of moles and R the universal gas constant.
 9-7 CLASSICAL THEORY OF SPECIFIC HEAT CAPACITY %1'7

9-7 CLASSICAL THEORY OF SPECIFIC HEAT CAPACITY

In thermodynamics, the change in internal energy U of a system, between two

equilibrium states, is defined by the equation

u.- u.- w•• ,

where w•• is the work in any adiabatic process between the states. Only changts
in internal energy are defined.
Starting with a molecular model of a system , we can identify the internal energy
with the sum of the energies of the individual molecules. In the preceding section
we have derived a theoretical expression for the total energy associated with the f
degrees of freedom of each of theN molecules o f a gas. We therefore set this equal
to the internal energy U:

(9-25)

The specific internal energy per mole is

u = !!_ =[RT. (9-26)

n 2
How can we tes t the validity of the assumptions made in the foregoing deriva-
tion 7 The most direct way is from measurements of specific heat capacities. The
molal specific heat capacity at constant volume is

c.=(;~)..
Hence, if the hypothesis above is correct, we should have

c.=..!!...(£
dT 2
RT) =f.R.
2
(9-27)

We also know from thermodynamic reasoning that for an ideal gas,

c" .. c. + R.
Hence

(9-28)
and
!+2
r=~ =-2- .. !+2. (9-29)
c. £ J
2
268 KINETIC THEORY !l-7

Table 9-1 Molal specific heat capacities of a number

of gases, at temperatures near room temperature.
The quantities measured experimentally are <1, andy.
The former is determined by use of a continuous
How calorimeter and the latter is obtained from
measuremeniS of the velocity of sound in the gas.

Ct•- C11
Gas y Cp/R c,/R -R-
- ------- - ----
He 1.66 2.50 1.506 .991
Ne 1.64 2.50 1.52 .975
A 1.67 2.51 1.507 1.005
Kr 1.69 2.49 1.48 1.01
Xe 1.67 2.50 I.SO 1.00
--- - -- -3.47
-- - 2.47
- - - -1.00- -
H, 1.40
o, 1.40 3.53 2.52 1.01
N, 1.40 3.50 2.51 1.00
co 1.42 3.50 2.50 1.00
NO 1.43 3.59 2.52 1.07
Cl, 1.36 4.07 3.00 1.07
------ - - - --- ----
co, 1.29 4.47 3.47 1.00
NH1 1.33 4.41 3.32 1.10
---
CH,
----
1.30
------
4.30 3.30
---
1.00
Air 1.40 3.50 2.50 1.00

Flg. 9-S A dumbbell molecule.

 CLASSICAL THEORY OF SPECIFIC HEAT CAPACITY 219

Thus while the principles of thermodynamics could give us only anlexpression

for the differtnce between the specific heat capacities at constant pressure and
constan t volume, molecular theory, together with the cquipartition principle,
predicts the actual magnitudes of the specific heat capacities a nd their ratio y, in
terms of the number of degrees of freedom f and the experimentally determined
universal constant R. Note that, according to the theory, c, , cp, and y arc all
constants independent of the temperature.
Consider first a gas whose atoms arc monato mic and for which the energy is
wholly kinetic energy of translation. Since there are three translational degrees of .
freedom,/ - 3, and we would expect

c• - [.2 R - 1
2 R - I .SR•

and

, .. ~ .. ~ - 1.67.
c. 3
This is in good agreement with the values of c, andy for the monatomic gases listed
in Table 9- l. Furthermore, the specific heat capacities of these gases arc found to
be practically independent of temperature, in agreement with the theory.
Consider next a diatomic molecule having the dumbbell structure shown in
Fig. 9-5. Its moment of inertia about the X· and z-axes is very much greater than
that about they-a xis, and if the latter can be neglected, the molecule has two
rotational degrees of freedom, the two quantities specifying the rotational kinetic
energy being the components of angular velocity about the x· and z-axes. Also,
since the atomic bond is not perfectly rigid, the atoms can vibrate along the line
joining them. This introduces two vibrational degrees of freedom, since the vi bra·
tiona! energy is part kinetic and part potential and is specified by the velocity and
the separation of the atoms. We might therefore expect seven degrees of freedom
for a diatomic molecule (3 for translation, 2 for rotation, and 2 for vibration).
Forf = 7, the theory predicts
7
c, = i R = 3.5R,

These values are not in good agr.ement with those observed for the diatomic
gases listed in Table 9-l. However, letting/"" 5, we find

c, = z5 R = 2.5R, 7
, .. 5 = 1.40.
 270 KINETIC THEORY 9-7

These are almost exactly equal to the average values of c. and y for the diatomic
molecules in the second part of the table (CI, is an interesting exception). Thus,
near room temperature, these molecules behave as if either the rotational or vibra-
tional degrees of freedom, but not both, shared equally with the translational
degrees of freedom in the total molecular energy.
As the number of atoms in a molecule increases, the number of degrees of
freedom cau be expected to increase; and the theory predicts a decreasing ratio
of specific heat capacities, in general agreement with experiment.
The main features of the theory are fairly well borne out. It predicts that y
is never greater than 1.67 or less than I and this is in fact true. However, if we insert
in Eq. (9-29) the measured values of y and solve for f, the result is in general not
exactly an integer. Now a molecule either has a degree of freedom or it has not.
Degrees of freedom are counted, not weighed. It is meaningless to speak of a
fraction of a degree of freedom, and the simple concept of equipartition is
certainly not the whole story.
When we examine the temperature variation of specific heat capacities, the
divergences between experiment and the simple theory above become even more
apparent. Except for gases whose atoms are monatomic, the specific heat capacities
of all gases increase with increasing temperature and decrease as the temperature
is lowered. In fact, at a temperature of20 K, the specific heat capacities of hydrogen
(the only diatomic gas that remains a gas at very low temperatures) decreases to
iR, the value predicated by theory for a monatomic gas. Thus at this low tem-
perature neither the rotational nor the vibrational degrees of freed om of the
hydrogen molecule appear to share at all in the change of internal energy associated
with a change in temperature. All of the difficulties mentioned above are removed,
however, when the principles of quantum mechanics and of statistics are taken
into consideration. These are discussed in Section 12-7.
The pressure of a gas depends on its translational kinetic energy, and regardless
of its molecular complexity a molecule has only three translational degrees of
freed om, and its translational kinetic energy equals 3kT/2. Then if u,, represents
this portion of the internal energy,

3
U., = i NkT.
The pressure P equals NkT/ V, so

(9-30)

where u., is the translational energy per unit volume, or the energy density; and, as
pointed out earlier, the pressure equals two-thirds of the translational energy
density.
 SPECifi C HEAT CAPACITY OF A SOLIO 271

f-1 SPECIFIC HEAT CAPACITY OF A SOLID

The molecules of a solid, unlike those of a gas, are constrained to oscillate about
fixed points by the relatively large forces exerted on them by other molecules. Let
us imagine that each executes harmonic motion. Each has three degrees of freedom,
considered as a mass point, but the por~ntlal energy associated with its motion,
which could be neglected for the widely separated molecules of a gas, is on the
average just equal to the kinetic energy, if the motion is simple harmonic. Hence,
if the equipanition principle is valid for solids, we must assign an energy kT toeach
degree of freedo m (kT/2 for kinetic energy, k T/2 for potential energy) rat her than
just kT/2 as for the molecules of a gas. The total energy of N molecules is then

U ~ 3NkT, (9-3 1)

and the molal specific heat capacity at constant volume, from the theory, is

c. • 3R = 3 X 8.31 X 10'- 24.9 X 10' J kilomole-1 K- 1• (9-32)

This is in agreement with the empirical law of Dulong and Petit which states
that at temperatures which are not too low, the molal specific heat capacities at
constant volume of all pure s ubstances in the solid state are very nearly equal to
3R. Again we have reasonably good agreement with experiment at high tempera-
tures. At low temperatures the agreement is definitely bad, since, as we have seen,
the specific heat capacities of all substances must approach zero as the temperature
approaches absolute zero. This is another problem to which the classical theory
does not provide the right answer a nd in which the methods of quantum mechanics
must be used.
One other d iscrepancy between simple theory and experiment should be pointed
out here. There is good reason to believe that in metals, which are electrical con-
ductors, each atom parts with one or more of its outer electrons and that these
electrons form a sort of electron cloud or electron gas, occupying the volume of the
metal and constrained by electrical forces at the metal surfaces in much the same
way that ordinary gases occupy a containing vessel. This electron gas has trans-
lational degrees of freedom which are quite independent of the metallic ions
forming the crystal lattice, and it should have a molal specific heat capacity equal
to that of any other monatomic gas, namely, 3Rf2. That is, as the temperpture of
the metal is increased, energy must be supplied to make the electrons move I'aste r as
well as to increase the amplitude of vibrations of the metallic ions. The latter
should have a specifi c heat capacity of 3R, so the total heat capacity of a metal
should be at least 3R + 3R/2 = 9Rf2. Actually, metals obey the Dulong-Petit law
as well as do nonconductors, so apparently the electrons do not share in the thermal
energy. This was a very puzzling thing for many years, but again it has a very
satisfact ory explanation when quantum methods are used.
 272 KINETIC THEORY

PROBLEMS

9-l (a) Compute the number of molecules per unit volume in a gas at 300 K when the
pressure is 10-.s Torr. (b) How many molecules are there in a cube of I mm on a side
under these condilions?
9-2 The model used in this chapter assumes that the molecules arc un iformly distributed
throughout the container. What must be the size of a cubical e leme nt of volume in the
container so that the number of particles in each volume element may vary by 0. 1 %w hen
the gas is at standard conditions? (From a study of statistics it can be shown that the
probable deviation of the number of particles in each volume element from the average
number of particles, N, is given by N''').
9-3 (a) In Fig. 9- l, let • - 45°, t.• - 0.01 radian, 0- 60°, and t.9- 0.0 1 radian.
What fraction of the molecules of a gas have velocity vectors within the narrow cone
which intercepts the shaded area t.A? (b) Consider a second cone intercep ting the some
orto on thelspherical surface, but for which 4o - 90°, 0 - 0. Sketch this cone and com-
pare the number of velocity vectors included within it with those in the cone of part (a).
9-4 (a) Approximately what fraction of the molecules of a gas have velocities for wh ich
the angle ¢ in Fig. 9-1 lies between 29.5° a nd 30.5°, while 0 lies between 44.5° and 45.5"?
(b) What fraction have speeds for which ¢lies between 29:5° and 30.5°, regardless of the
value 9? [Note: Angles must be expressed in radians.)
9-S Suppose that the number of molecules in a gas having speeds between v and v + t.v
is given by t.N, - N llv/v0 for v0 > v > 0 and t.N, - 0 for v > v0 • (a) Find the frac-
tion of molecul es having speeds between 0.50 v0 and 0.5 1 v0 • (b) Find the fraction having
the speeds in part (a) in the direction described in part (a) and part (b) of the previous
problem. (c) Find the flux of molecules described in part (b) of this problem arriving at a
surface, if the gas is at standard conditions.
9-6 Calculateliand ••m• for the following distributions of six particles: (a) all have speeds
of 20m s- •; (b) th ree have speeds of 5 m s-• and three have speeds of 20 m s-•; (c) four
have speeds of 5 m s-• and two have speeds of 20m s-•; (d) three are at rest and three
have speeds of 20m s-•; (e) one has a speed of S m s-•, two have speeds of 7 m s-1, two
have speeds of 15 m s- • and one has a speed of 20m s- •.
9-1 T he speed distribution function of a group of N particles is given by t.N, - kv au
for v0 > v > 0 and t.N, - 0 for v > v0 • (a) Draw a graph of the distribution function.
(b) Show that the constant k - 2N/<l (c) Compute the average speed of the particles.
(d) Compute the root-mean-square speed of the particles.
9- 8 (a) Derive Eq. (9-7) beginning with Eq. (9-4). (b) For a gas at standard conditions,
find t.<l>, for molecules obeying the speed distribut ion law of the previous problem and
having speeds between O.SOv0 and O.SI v0 • (c) Determine <I> for molecules having the same
speed distribution.
9-9 What form would Eq. (9-17) take if several kinds of molecules were present in a gas?
Does the answer agree with Dalton's law?
9- JO Derive a n expression equivalent to Eq. (9- 17) for a two-d imensional gas, i.e.,
one whose molecules can move only in a plane. (The concept corresponding to pressure,
o r force per unit area, becomes force per uni t lengt h.)
 PROBLEMS 273

9-11 (a) Compule lhe rms speed of a gas of helium aroms a l 300 K. (b) AI whal lem-
peraiUre will oxygen molecules have rhe same rms speed? (c) Through whar polenrial
difference musr a singly ionized oxygen molecule be acceleraled 10 have I he same speed?
9- ll (a) How many molecular impacrs are made per second on each sq uare cenrimeler
of a surface exposed 10 air al a pressure of I arm and al 300 K? The mean molecular
weight of air is 29. (b) Whal would be the lengrh of a cylinder I em' in cross seclion con-
raining rhe number of air molecules ar I arm and 300 K which collide wilh a svrface I em•
in one second? I
9-q A cubical box of 0. 1 m on a side conrains 3 x totS molecules of 0 1 ar 300 K.
(a) On rhe average, how many collisions does each molecule make wirh rhe walls of I he
box in one second? (b) Whar pressure does I he oxygen exerl on rhc walls of I he box?
9-14 A closed vessel conrains liquid wa rer in equilibrium wirh its vapor al too•c and
I arm. One gram of warer vapor al I his rernperarure and pressure occupies a volume of
1670cm1 • The heal of vaporizarion ar lhis remperarure is 22501 g-1• (a) How many
molecules are there per em' of vapor? (b) How many vapor molecules mike each em'
of liquid surface per second? (c) If each molecule which srrikcs rhc surface condenses,
how many evaporare from each em' per second? (d) Compare lhe mean kineric energy
of a vapor molecule wirh rhe energy required 10 rransfer one molecule from rhe liquid 10
rhe vapor phase.
9- 15 When a liquid and irs vapor are in equilibrium, lhe rales of evaporarion of lhe
liquid and condensarion of rhe vapor are equal. Assume rhal every molecule of rhe vapor
srrik.ing rhe liquid surface condenses, and assume rhar rhe rare of evaporarion is rhe same
when rhe vapor is rapidly pumped away from rhe surface, as when liquid and vapor arc
in equilibrium. The vapor pressure of mercury al o•c is ISS x J0-4Torr and rhe laren t
hear of vaporizarion is abour 340 J g-•. Com pure rhe rare of evaporarion of mercury inro
a vacuum, in g em-• s-•, ar a remperarure of (a) o•c, (b) 20°C.
9-16 A rhin-walled vessel of volume V conrains N parriclcs which slowly leak our of a
small hole of area A. No parricles enler rhe volume lhrough rhe hole. Find I he rime
required for rhe number of parlicles 10 decrease lo N/2. Express your answer in rerms of
A, V, and tl.
9-17 The pressure in a vacuum sysrem is ro-• Torr. The exrernal pressure is I a rm and
T - 300 K. There is a pinhole in rhe walls of rhe sysrem, of area ro- 10 em•. Assume rhal
every molecule "srriking" Ihe hole passes lhrough ir. (a) How many molecules leak inlo
rhe sysrem in I hou r ? (b) If rhe volume of rhe sysrem is 2 lirers, whal rise in pressure
would resull in rhe sysrem? (c) Show rhar rhe number of molecules rhat leak our is
negligible.
9-18 A vessel of volume 2 V is divided inro comparr menrs of equal volume by a lhin
parririon. The lefl side conrains imrially an ideal gas ar a pressure P0 , and rhe righl side is
initially evacuated. A small hole of area A is punched in I he partition. Derive an expression
for rhe pressurcP1 on rhe lefl side as a funcrion of rime. Assume rhc remperarure 10 remain
conslanr and 10 be rhe same on borh sides of rhe parririon.
9-19 An insulaled chamber conlaining liquid helium in equilibrium wirh irs vapor is
mainrained ar 1.2 K. II is separared from a second insula red chamber mainrained ar
300 K , by a rhin insularing parririon wirh a small hole in il. The helium vapor is allowed
ro fill borb chambers. If rho vapor pressure of rhe hel ium ar 1.2 K is P0 , show rhar rhe
 274 KINETIC THEORY

pressure Pin the other chamber is P0 VJ00/1.2. (The ratio of P/P0 is called the thtrmo-
moltcufar pr~ssur~ ratio and is important in vapor pressure thermometry when the pressure
is so low that the panicles do not collide in a distance long compared to linear dimensions
of the apparatus.)
9- 20 An ideal monatomic gas is confined to an insulaied cylinder fiued wit h an insulated
piston. (a) By considering collisions of the m~lecules or the gas with the quasistatically
moving piston, show thai PVIi' - constant. (b) Determine the pressure dependence o f
the rms speed of the molecules in an adiabalic compression or expansion.
9-21 A molecule consists of four atoms at the corners of a tetrahedron. (a) What is the
number or translational, rotational, and vibra tiona l degrees or freedom for this molecule?
(b) On the basis of the equiparrilion principle, what are the values of c. and y for a gas
composed or these molecules?
9-lZ Under the action of •uitable radiation a diatomic molecule splits into two atoms.
The ratio of the number or d issociated molecules ro the total number or molecules is «.
Plot y( - c1./c.) as a function of"' al a temperature where the vibrational modes of the
diatomic molecule are excited.
9-23 F ind the total translational kinetic energy and the rms speed of the molecules of 10
liters of helium gas at an equilibtium pressure of 10' N m-•.
9-24 (a) Find the specific heat capacity at constant volume for a gas of H1 molecules and
H 10 molceules. (b) How do the specific heat capacities cha nge if the gas is liquefied or
solidified?
 10
Intermolecular forces.
Transport phenomena
10- 1 INTERMOLECULAR FORCES

10-2 THE VAN DER WAALS EQUATION OF STATE

10- 3 COLLISION CROSS SECTION. MEAN FREE PATH

1<l-4 COEFFICIENT OF VISCOSITY

10-5 THERMAL CONDUCTIVITY

10-6 OIFFUSION

10-7 SUMMARY
 276 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA 1 ~2

1~1 INTERMOLECULAR FORCES

Io the preceding chapter, the molecules of a gas were treated as geometrical points
that e><erted no forces on each other. We now wish to take such forces into account.
The force between any pair of molecules is of electrical origin; and because
o f the complicated structure of an atom or molecule, it is not expressible by any
simple law. In general, at relatively large separations, the force is one of attraction,
referred to as a van der Waals force, which decreases rapidly wlth increasing
separation.! When two molecules approach so closely that their electron clouds
overlap, the force becomes one of repulsion that rises very rapidly as the separation
becomes smaller. Thus the intermolecular force must have the general form of the
solid curve in Fig. 1(}...1.

Fl&. ID-1 Intermolecular forces.

The simplest approximation to this law is to treat the molecules as elastic hard
spMres, for which the force of repulsion becomes infinite when the surfaces of the
spheres come into contact. If we include a force of atlraction when the molecules
are not in contact, the force law has the form of the dolled curve in Fig. J(}...l.

1 ~2 THE VAN DER WAALS EQUATION OF STATE

We have made extensive use of the van der Waals equation of state in earlier
chapters, not so much because o f any great accuracy of this equation in describing
the properties of real gases but because it shows in a general way, through the
factor a, how these properties depend on intermolecular forces of auraction, and
through the factor b how they depend on molecular sizes.
 10-2 THE VAN DER WAALS EQUATION OF STATE 277

The latter correction to the equation of state was actually first suggested by
Clausius. He reasoned that in the derivation in Section 9-4 one should use not the
actual volume V of the container, but the volume available to a single molecule,
which will be somewhat less than V because of the volume occupied by the other
molecules. If we represent the "unavailable" volume per mole by b, then in a gas
consisting of n moles the unavailable volume is nb and we should write
P(V - nb) = nRT,
or, dividing th rough by n,
P(v- b)~ RT. (10-1)
This eq uation was fi(St written down by Hirn. • (Here, the letter v represents the
molal specific volume, not the molecular speed.)
.,-_- -;- ......
,/."~~.1'~ ' ,

:.;;.r.. .
I t,
,...,. \-,;.. I··,....... I
''~l~P~ I
I I
I-i-I
Fig. 10-2 The radius
of the sphere of exclu-
sion equals the molec-
ular diameter d.
If the molecules are considered as hard spheres of diameter d, the minimum
distance between the centers of two molecules, as shown in Fig. 10-2, is equal to d.
In effect, the center of each molecule is excluded by the other from a sphere of
radius d, known as the "sphere of exclusion." The volume of this sphere is 4-trd0f3,
and to avoid counting each pair twice, we take as the tot:.l unavailable volume, for
a system of N molecules,

The number of molecules N is the product of the numbe r of moles nand Avo-
gadro's number N... so the unavailable volume per mole, or the constant b, is

b =~ N._.,d•. (10-2)
3
This is four times as great as the actual molecular volume per mole, which is
I
(,N...,d'. ( 10-3)

Vander Waals, in 1873, included a second correction term in the equation of

state to take into account the force of attraction between molecules. Let us assume
• Gustav A. Hirn, French engineer, (1815-1890).
 278 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA 10-2

that these forces decrease so rapidly with distance (for example, as 1/r') that they
are appreciable only between a molecule and its nearest neighbors. Molecules
within the body of the gas are on the average attracted equally in all directions, but
those in the outermost layers experience a net inward force . A molecule approaching
the wall of the contain:r is therefore slowed down and the average force exerted
on the wall, and hence the observed pressure, is somewhat smaller than it would
be in the absence of attractive forces.
The reduction in pressure will be proportional both to the number of molecules
per unit volume in the outer layer, n "' N/V, and to the number per unit volume
in the next layer beneath them, which is doing the attracting. Hence the pressure
will be reduced by an amount proportional to n1, o r equal to ""'• where « is a
factor dependent on the strength of the attractive force . Since the number of
molecules N equals nN•• where n is the number of moles, then
«N~ -a
1
«D
1
- «(N)' 1 n
- =«N.-;;;;--- (10-4)
v v•" ,l'
where the product «N! bas been replaced by a. Thus the pressure Pgiven by tbe
Hirn equation,
p- ..B!....
v-b
should be reduced by afv' ; and
p ~ RT -~
v - b v•'
or
(P + ;.)(o - b)- RT, (1(}...5)

which is the van der Waals equation of state.

Since the molal specific critical volume of a van der Waals gas, v., is equal to
3b, it follows from Eq. (1(}...2) that
1
00 • 3b = 2N._1Td , (10-6)
which is 12 times the total molecular volume. The value of b for a van der Waals
gas therefore provides a means of estimating molecular diameters, since

d= (2£._)"·.
211N._
(1(}...7)

Thus for helium, for which b ~ 23.4 x l<r' m' kilomole- 1 , we have

d = (
3 X23.4 X104 )Ill~ 2.6 X 10- ••m • 2.6 x 1<Jicm.
2 X 3.14 X 6.02 X 1010
Other methods of estimating molecular diameters will be described in Section
1(}...4. Values of a and b for several gases are given in Table 2- 1.
 10-3 COLUSION CROSS SECTION. MEAN FREE PATH 279

10-3 COLLISION CROSS SECTION. MEAN FREE PATH

In deriving the expression for the pressure exerted by a gas, the molecules were
treated as geometrical points which could fly freely from one wall of a container
to the other without colliding with other molecules. O ne of the objections raised in
the early development of kinetic theory was that if molecules acted in t his way, a
small amount of gas released in a large room would spread throughout the room
practically instantaneously, whereas we know that when the stopper is removed
from a bottle of perfume, a considerable time elapses before the odor can be
detected even at a point only a few feet away, in lhe absence of air currenls. It
was soon realized that this relatively slow diffusion of one gas in another resulted
from molecular collisions such as that shown in Fig. 10-3, which cause a molecule
to move in an irregular, zigzag path.

0
0 0

0
0
0

Fig. J0-3 Molecular free paths.

We again assume lhat a molecule is a hard sphere. Let us refer to one of the
colliding molecules as the " target" molecule and to lhe other as the "bullet"
molecule. Then a collision occurs whenever the distance between the centers of
the molecules becomes equal to the molecular diameter d, as in Fig. 10-2.
Since it is only the center-to-center distance that determines a collision, it does
not matter whether the target is large and the bullet small, or vice versa. We may
therefore consider the bullet molecule to shrink to a point at its center, and the
target molecule to occupy the entire sphere of exclusion , of radius d.
Now consider a thin layer of gas of dimensions L, L, and tix, as in Fig. 10-4.
The layer contains (equivalent) target molecules, represented by the shaded circles.
We then imagine that a very large number N of bullet molecules, represented by the
black dots, is projected toward the face of the layer-like pellets from a shotgun-
in such a way that they are distributed at random over the face of the layer. If the
thickness of the layer is so small that no target molecule can hide behind another,
the layer p resents to the bullet molecules the appearance of Fig. 10-4.
 280 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA

.e
,
·- : ..
d.
Fig. 10-4 A thin layer of gas of
"larget" molecules being bombarded
by "bullet" molecules.

Most of the bullet molecules will pass th rough the layer, but some will co!tide
with target molecules. The ratio of the number of collisions, t.N, to the total
number of bullet molecules, N, is equal to the ratio of the area presented by the
target molecules to the total area presented by the layer:
fiN target area
N = total area ·
The target area a of a single (equivalent) molecule is the area of a circle of
radius d, the exclusion radius:
a= .,.d'. (10-8)
This area is called the microscopic collision cross stction of one (equivalent) mole-
cule. The total target area is the product of this and the number of target molecules
in the layer. If there are n target molecules per unit volume, this number is nL' t:.x,
so the total target area is
naL'fix.
The total area of the layer is L', so

fiN = nal! fix = na fix. (10-9)

N L'
The quantity na is called the macroscopic collision cross stction of the
(equivalent) molecules. Since the number density n, in the M KS system, is the
numbe r of molecules per cubic meter and the collision cross section a is the number
of square meters per molecule, the unit of the product na is I square meter per cubic
meter (I m• m-> = I m- 1). More generally, in any system, the unit of macro-
scopic collision cross section is a rtciprocalltngth, not an area.
 10-3 COLLI SION CROSS SECTION. MEAN FREE PATH 211

Each of the AN collisions diverts a molecule from its original path or scauers
it out of the beam, and decreases the number remaining in the beam. Let us,
therefore, interpret AN not as a "number of collisions," but as the decuase in the
number N, and write
AN ~ -NnaAx,
or
AN= -na Ax.
N
In reality, N decreases in stepwise fashion as individual molecules make
collisions, but if N is very large we can consider it a continuous function of x and
write
dN = - nadx.
N
Then
InN= -nax +constant;
and if N = N 0 , when x = 0,
N = N 0 exp(- nax). (10-10)
This is known as the survival tquation. It represents the number of mollcules N,
o ut of an initial number N0 , that has not yet made a collision after traveling a
distance x.
Inserting the expression for N in Eq. (10-9), we obtain
AN= N 0naexp(-nax)Ax. (10-11)
In this equation, N is the number of molecules making their first collision after
having traveled a distance between x and' x + Ax.
Let us calculate the average distance traveled by a group of N0 molecules
before they make their first collision. This average distance is known as the mean
fru patlr, /, To calculate it, we multiply x by the number of particles AN that
travels the distance x before colliding, sum over all values of x, and divide by the
total number N0 • Replacing the sum by an integral, we have

1 ~ _;~
:.._x__
No •
f."
AN = na x exp( - nax) dx .

The definite integral equals 1/n' a', so

1-..!... (10-12)
na
and the mean free path is inversely proportional to the macroscopic collision cross
section. Since the unit of macroscopic collision cron section is the reciproc•l of the
unit of length, the unit of mean free path is the unit o f length. Note that the mean
free path does not depend on the molecular speed.
 282 INTERMOLECULAR FORCES. TRANSPORT PHENOMEN A 10-3

The concept of mean free path may be visualized by thinking of a man shooting
bullets aimlessly into a thick forest. All of the bullets will eventually hit trees, but
some will travel farther than others. It is easy to see that the average distance
traveled will depend inversely on both the denseness of the woods (n) and on the
size of the trees (a).
A common experimental technique is to project into a gas a beam of particles
(either neutral or having an electric charge) and to measure the quantity N0 and
the number N remaining in the beam after a distance x. The exponential decrease
predicted by Eq. ( 10-10) is found to be well obeyed, and we may now reverse the
reasoning by which this equation was derived. That is, since N 0 , N, and x are all
measurable experimentally, Eq. (10-10) can be solved for na or /, and we can
consider these quantities to be defined by Eq. (10-10), q uite independently of any
theory of molecular collisions.
Although we de rived the equations above by considering a beam of molecules
projected into a gas, the mean free path is the same if the group is considered to
consist of the molecules of a gas moving at random among the other molecules
and making collisions with them. The motion of a single molecule is then a zigzag
path as suggested in Fig. 10-3, and we can understand why it is that although
the average molecular speed is very large, a molecule wanders away fro m a given
position only relatively slowly.
As an example, suppose the molecular diameter d equals 2 X m. At JO-••
standard conditions, there are about 3 x 10" molecules m- 3 in a gas. T he macro-
scopic collision cross section is then
DC1 = n1rd'"" 3 X 1025 X 3.14 X 4 X J0-2<1"" 40 X J()' m-1,,
and the mean free path is
I = ..!. "" 2.5 x JO-' m,
na
which is smaller than the wavelength of visible light. The average intermolecular
separation at standard conditions is about 3 x 10-• m, so the mean fr~e path is
much la rge~ than the average intermolecular separation, and Fig. 10-3 is there-
fore misleading.
Since the number of molecules per unit volume. n, is inversely p roportional
to the pressure, the mean free path increases as the pressure is decreased. A moder-
ately good "vacuum" system will reduce the pressure to 1()-3 Torr, which is about
tO- • atm. The mean free path is then a million times that at atmospheric pressure,
or of the order of 25 em.
More complete theories of the mean free path take into account the relative
motion of a lithe molecules of a gas, that is, they consider the " target" molecules,
as well as the "bullet" molecules, to be in motion. The only change in the end
result is to introduce a small correction factor in Eq. ( 10-12). The inverse depen-
dence on the number of molecules per unit volume and on the collision cross
 10-3 COLLISION CROSS SECTION. MEAN FREE PATH 283

section remains unaltered. On the assumption that all molecules have the same
speed, Clausius obtained the result
3 I 0.75
1=- - = - .
4 na na
If the molecules have a Maxwellian velocity distribution (see Section 12-2),
I = __!_,..!. = 0.707 .
.j2 na na
However, we shall continue to use the simpler result of Eq. (10-12).
In the preceding discussion, the target molecules and bullet molecules were
considered identical hard spheres, each of diameler d. One often wishes to know
the mean free path of an electron, moving among the neutral or ionized molecules
of a gas in a plasma, or among the fixed metallic ions in a metallic conductor. The
"diameter" of an electron is so much smaller than that of a molecule that the elec·
tron can be considered a geometrical point, and the center-to-center distance in a
collision (see Fig. 10-2) becomes d/2 rather than d, where d is the molecular
diameter. Furthermore, the velocities o f the electrons are so much greater than
those of the molecules that the latter can be assumed at rest, and the correction
for relative velocities need not be made. From the considerations above, the
electronic mean free path 10 is
lo = 4..!., (10-13)
na
where n is the number density of molecules and na is the macroscopic collision cross
section of electrons with molecules or ions.
In terms of the mean free path, the survival equation can be written

N = N 0 exp(-nax) = N 0 exp(-x/l). (10-14)

Figure 10-5 is a graph of this equation, ir. which the dimensionless ratio N/N0 is
plotted as a function of xfl. The ordinate of the curve is the fractional number of
molecules with free paths longer than any fraction of the mean free path. Note that
the fraction with free paths longer than the mean is exp (-I) or 37%. while the
number with free paths shorter than the mean is 63 %.
An interesting aspect of the theory of distribution of free paths is that the N0
molecules considered originally are not necessarily just starting out on their free
paths after having made a collision. We merely make a random selection of a
large number of molecules at any instant and inquire into their future without
asking questions about their past. Sometimes, however, it is the past rather than
the future that is of interest. That is, we may fix our attention on a group of
molecules at some instant and instead of asking, as we did above, how far each
will travel on the average before it makes its next collision, ask how far each has
traveled on the average since making its last previous collision. The same reasoning
 284 INTERM OLECULAR FORCES. TRANSPORT PHENOMENA 10-3

as that above shows that this average distance is also the free path /, and that the
distribution of "past" free paths is Ihe same as the dis tribution of "future" free
paths. Hence when we consider a large number of molecules in a gas at any instant,
the average dista nce they ha ve yet to travel before their next collision is equal to
the average d istance they have already traveled since their last collisions, and both
distances are equal to the mean free path /. We shall make use of this fact in the next
section, in calculating the average distance above or below a plane at which mole-·
cules make their last collision before crossing the plane.

Fla. 10-s Graph of 1he survival

equation.

This result raised the following interesting question. If the average distance
traveled by the group bifou we conside r it is /, and the average distance ofiu we
consider it is also /, why is the mean free palh not equal 10 21 ralher !han I?
Another important concepl is !hal of rollisionfr~qumq z, !he average number
of collisions per unit l ime made by a molecule wilh other molecules. In a time
inlerval At, a molecule !ravels an average dislance;; Ar along irs zigzag palh. The
average number of collisions il makes in !his lime is ii At{/, and hence !he collision
frequency is

z c:-
ii = -unu. (10-15)
I
From the values of ii, n, and a for oxygen molecules at room temperature, we find

z ,., 5.5 x 101 collisions s·•.

The mtanfite timeT, or the average time berween collisions, is the reciprocal
of the collision frequency z and hence
T= ! =~=...!..., ( IQ-J6)
z ii Vna
 !G-3 COLLISION CROSS SECTION. MEAN FREE PATH 285

For oxygen molecules at room temperature,

T "' _ _ I - , "' 1.8 X 10-10 S.

5.5 X 10
The preceding results form the basis of the theory of me tallic conduction de-
veloped by Drude* in 1900. We assume tha t the free electrons in a metallic con-
d uctor can be considered an ideal gas and that their average random speed ii is the
same as that of gas molecules of the same mass, a t the same temperature. (We shall
sho~ in Chapter 13 tha t this is not a very good assumption). If the electric field
intensity in the conductor is£, the force F on each electron, of (negative) charge
e, is F~ tE. As a result of this force, the electrons have an acceleration a opposite
to the direction of the field and of magnitude
F e£
a= - = - .
m m
The electrons do not accelerate indefinitely, however, because of collisions with
the fixed metallic ions. We assume tha t at each such collision a n electron is brought
to rest and makes a fresh start losing a ll memory o f its previous velocity. In the
mean free timeT between collisions, an electro n acqui res a velocity opposite to the
field equal to aT, and its average velocity between collisions, or the drift velocity
u , is
U = ! QT = !(~)!!,
2 2 m ii
This drift velocity is superposed on the ra ndom "thermal" velocity ,li ~ut in a n
actual conductor it is very small compared with the random velocity. Note that
in the expression for the mea n free path 1•• we should use Eq. (10- 13).
The current density J in the metal (the current per unit of cross-sectional a rea)
is the product of the num ber density n,. of electrons, their cha rgee, and the drift
velocity u:

J = n,eu - (-n,•'
-'•) E.
2mv
The resistivity p of the metal is defined as the ratio of the electric intensity E
to the current density J : p = EfJ. Hence
2mii
( 10-17)
p = n.•''·.
I n a given metal a t a give n te mpe rature, all quantities on the right side of the
preceding equation a re consta nts so tha t !he Drude theory predicts !hat under lhese
condirions !he resisrivity of a metallic conductor is a consta nt independent of E.

• Paul K. L. Drude, German physicist (l863- t906).

 288 INTERMOLECULAR FO RCES, TRANSPORT PHENOMENA 10-4

In other words, the current density J is directly proportional to the electric inte nsity
E, and the metal, in agreement with experiment, obeys Ohm's Jaw.

A more familiar statement of Ohm's law is that at a given temperature, the

potential difference V between two points of a conducting wire is directly proportional
to the current/ in the wire, or that V - IR, where R is a constant independent of I.
The total current I in a conductor of constant cross-sectional area A is I = JA. If
the length of the conductor is L, the potential difference Vbetween its ends is V - EL,
so the equatioo pj - E can be written

or,

where the rtJistance R - pL/A.

v
We shall show in Chapter 12 that the average rando m velocity in a gas is
proportional to T'", so the theory predicts that the resistivity p should increase
with the square root of the temperature. However, experimentally the resistance
of metallic conductors increases linearly with increasing temperature, so the Drude
theory is far fro m complete.

10-4 COEfFICIENT OF VISCOSITY

In the next three sections, we give an elementary treatment of three properties of a

. gas described by the general term o r transport phenomena. These are its viscosity,
thermal conductivity, and coefficient of diffusion, and they can be explained in
terms of the transport across some imagined surface within the gas of momentum,
energy, and mass, respectively. Consider fi rst the coefficient o r viscosity.

Fig. HHi Viscous flow between a stationary

lower plate and a moving upper plate.
 10-4 COEFFICIENT OF VISCOSITY 287

It appears contradictory at first sight that a gas consisting of widely separated

molecules making perfectly elastic collisions with one another should exhibit any
viscosity or internal friction. Every real gas, however, is viscous; and we now show
that Ihis property is another necessary consequence of our simple model and does
nol require the assignment of any new properties to the molecules.
Figure 10-6 represents a portion of two large plates separated by a layer of
gas of thickness L . Because of !he viscosity of the gas, a force F must be exerted
on the upper plate to d rag it to the right at constant velocity relati ve to the lower,
stationary plate. (An equal and opposite force must be exerted on the lower
plate to keep it at rest.) The molecules in the layer of gas have a forward velocity
component u which increases uniformly wich the distance y above the lower p late.
The co~ffici~nt of viscosity of the gas, '1· is defined by the equation

(10-18)

where A is the area of either plate and dufdy is the v~locity gradient at right angles
to the plates.
In the MKS system, the unit ofF/A is I newton per square meter and the unit
of the velocity gradient dufdy is I meter per second, per meier. The unit of the
coefficient of viscosity '1 is cherefore I newton per square meter, per meter per
second per meter, which reduces to I N s m-•. The corresponding cgs unit is
I dynes cm-•and is called I poise in honor of Poiseuille. • (I poise - 10 N s m-1.)
The forward velocily u of the molecules is superposed on their large random
velocities, so chat the gas is not in thermodynamic equilibrium. However, in most
praciical problems the random velocities are so much larger than any forward
velocity that we ca h use !he results previously derived for an equilibrium state.
The dotted line S·S in Fig. 10-6 represenls an imagined surface within the gas
at an arbicrary height y above the lower plate. Because of their random motions,
there is a molecular flux <I> across the dotted surface, bolh from above and from
below. We shall assume that at its last collision before crossing the surface, each
molecule acquires a flow velocicy toward the right, corresponding to the particular
height at which the collision was made. Since che flow velocity above the dotted
surface is greater than that below the surface, molecules crossing from above
transport a greater momentum (toward the right) across the surface than do the
molecules crossing from below. There results a net rate of transport of momentum
across the surface, and from Newton's second law we can equate the net rate of
transport of momentum, per unit area, to the viscous force per unit area.
Thus the viscosicy of a gas arises not from any "frictional" forces between its
molecules, but from the fact that they carry momentum across a surface as a result

*Jean-Louis M. Poiseuille, French physician (1799-1869).

 2a8 INTERMOLECULAR FOR CES. TRANSPORT PHENOMENA 1()-4

of their random motion. The process is analogous to that of two freight trains of
open-top coal cars moving in the same direction on parallel tracks at slightly
different speeds, with a gang of laborers in each car, each laborer shoveling coal
from bis car into the opposite car on the other track. The cars in the slower train
are continually being struck by pieces of coal traveling slightly faster than the cars,
with the result that there is a net forward force on that train. Conversely, there is a
net backward force on the faster train, and the effect is the same as if the sides of the
cars were rubbing together and exerting forces on one another through the
mechanism of sliding friction.

Fls. JG-7 The last mean free path

before the molecule crosses the surface
started a distance y - I cos 0 from the
surface.

Let us compute the average height j above (or below) the surface at which a
molecule made its last collision before crossing. In Section 9-3, we assumed that
the molecules were geometrical points and that all O<fou-molecules in the slant cylinder
of Fig. 9-2 would arrive at the area 6A without having made a collision. This
cannot be correct, because on the average each molecule travels only a distance I
without colliding with another molecule. These molecular collisions will not affect
the total ftux of 6<,6t>-molecules arriving at the surface, because for every collision
that scatters a B.Pt>-molecule out of the number originally in the cylinder, there
will be another collision that results in an identical O<Pt>-molecule at essentially the
same point. However, as explained in the preceding section, molecules arriving at
the surface will on the average have started their last free paths before reaching the 1
surface at a distance I away from the surface. The perfHndicular distance y from the
surface, for any 6-molecule (see Fig. 10-7) is y = I cos 0. The average value of y,
or j, is found by multiplying I cos 0 by the flux 6<1>1 , summing over all values of 0,
 COEFFICIEN T OF VISCOSITY 289

and dividing by the total flux ell. From Eq. (9-6), replaci ng :Evn,. with iio,

t.ell, = ~ iio sin 6 cos 6 t.O;

and from Eq. (9-11),
I
ell ~
4vn.
Therefore, replaci ng t.O with dO and integrating over 6 from zero to '"/2,
1 iiniJ;'' sin 6 cos1 6 dO
j = 2 -~I. (IQ-19)
-1 -vn 3
4
Hence on the average, a molecule crossing the surface makes its last collision
before crossing at a distance equal to two-thirds of a mean free path above (or
below) the surface.
Let u0 represent the forward velocity o f the gas a t the plane S-S. At a d istance
21/3 above the surface, the forwa rd velocity is
u=u0 +~I~
3 dy.
since the forward velocity gradient dufdy can be considered constant over a distance
of the order ofa free path. The forward momentum of a molecule with this velocity
is

mu = m(u• + 3~ 1~)-
dy
Hence the net momentum C:! in the direction of flow, carried across the surface
per unit time a nd per unit area by the molecules crossing from above, is the product
of the momentum mu and the total flux ell:

c'i!- lnmii(uo + ~ 1 ~)-

4 3 dy
Similarly, the momentum carried across the surface by the molecules crossing
from below is
,. f • -I num
v
_ (u -
2 d u) .
- I-
0
4 3 dy
The net rate of transport of momentum per unit a rea is the d ifference between
these quantities, or
c'i = l nmiil~ (IQ-20)
3 dy.
290 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA 10-4

700 IN<
I
I
600 I
I A
I
I
500
I
I I
I .~ Ht

..
.3<400
I
I ./ ./
300 /
I
/
/
,/ ·
;w /
100
10 IS lO 2S J() 3S
,/T(K'"' )

Fig. 10-8 The viscosity of helium, argon, and

neon is almosl a linear function of ,fi'.

Table 10-1 Values of the mean free path and molecula r diameter of
some gases determined from viscosity measurements. The values of
I and din this table were calculated using Eq. (10-13) for I.

~ o5•q I (I 5•c. 1 atm) d

Gas (N s m-•) (m) (m)

He 19.4 x w-• 18.6 x 1o-• 2.18 X lo-10

Ne 31.0 13.2 2.60
A 22.0 6.66 3.64
H, 8.71 11.8 2.74
N, 17.3 6.28 3.76
o, 20.0 6.79 3.60
co, 14.5 4.19 4.60
n
NH1 9.7 4.51 4.44
CH4 10.8 5.16 4. 14
c
n
 IC>-4 COEFFICIENT OF VISCOSITY 281

and from Newton's second law this equals the viscous force per unit area. Hence,
by comparison with the definition of the coefficient of viscosity in Eq. (IQ-18), we
have I
{IQ-21)

An unexpected conclusion from this equation is that the viscosity of a gas is

independent of the pressure or density, and is a function of temperature alone
through the dependence of ii on T. Experiment bears this out, however, except at
very low pressures where the mean free path becomes of the order of the dimen-
sions of the apparatus. The theory above would not be expected to hold under these
conditions, where a molecule could go bouncing from o ne wall to the other without
making a large number of collisions on the way. ·
We shall show in Section 12-2 that the mean speed ii is given by

ii= ,;lfkT.
; -;;;
so that

(10--22)

Thus for molecules of a given species, the theory predicts that '1 is proportional
to .Jr. and that for different species at a given temperature it is proportional to
.J;.fa.
Figure JQ-8 shows some experimental values of the viscosities of helium, neon,
and argon, plotted as functions of JT. The graphs are very nearly straight lines,
but they curve upward slightly, indicating that the viscosity increases with tem-
perature at a somewhat greater rate than predicted by the "hard-sphere" theory.
This can be explained by realizing that the molecules are not truly rigid spheres and
that a "collision" is more like that between two soft tennis balls than between two
billiard balls. The higher the temperature, the greater the averuge molecular kinetic
energy and the more the molecules become "squashed" in a collision. Thus the
center-to-center distance in a collision, and the corresponding collision cross
section a, will be slightly s maller, the higher the temperature, with a corresponding
increase in '1·
As for the dependence of viscosity o n the cross section a, Eq. ( IQ-22) is as a
matter of fact one of the relations used to " measure" collision cross sections and the
corresponding hard-sphere diameters d. Some values of d computed from viscosity
measurements, are given in Table 10--1.
 292 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA 10-5

10-5 THERMAL CONDUCTIVITY

The thermal conductivity of a gas is treated in the same way as its viscosity. Let
the upper and lower plates in Fig. HH) be at rest but at different temperatures, so
that there is a temptrature gradient rather than a velocity gradient in the gas. (It
is difficult to prevent conductive heat flow in a gas from being masked by convection
currents. The gas layer must be thin, and the upper plate must be a t a higher te m-
perature than the lower.) If dTfdy is the temperature gradient normal to a surface
within the gas, the thermal conductivity A is defined by the equation

dT
H =-A dy , ( 1(}...23)

where His the heat flow or heat current per unit area and per unit time across the
~urface. The negative sign is included because if dTfdy is positive the heat current
is downward and is negative.
In the MKS system, the unit of His I joule per square meter per second and
the unit of the temperature gradient dT/dy is I kelvin per meter. The unit of
thermal conductivity A is therefore I joule per square meter per second , per kelvin
per meter, which reduces to I J m- • s- • K - 1•
From the molecular viewpoint, we consider the thermal conductivity of a gas
to result from the net flux of molecular kinetic energy across a surface. The total
kinetic energy per mole of the molecules of an ideal gas is simply its internal
energy u, which in turn equals c, T. The average kinetic energy of a single molecule
is therefore c.Tdivided by Avogadro's number, N,, and if we .jefine a " molecular
heat capacity" c: as c: = cJN,., the ave rage molecular kinetic energy is c:r.
We assume as before that each molecule crossing the surface made its last
collision at a distance 21/3 above or below the surface. and that its kinetic energy
corresponds to the temperature at that distance. If T0 is the temperature at the
surface S-S, the kinetic energy of a molecule at a distance 21/3 below the s urface is

The energy transported in an upward di rection, per unit area and per unit
ti me, is the product of this quantity and the molecular flux <1>:

H• = 4! niic*(, T. -
I O
?:.1
3 dy
dT)·
In the same way, the energy transported by molecules crossing from above is

H l = -I nvc•
4
_ ( T.
• 0 + -32 I -dT)
dy
.
 t0-5 TH ERMAL CONDUCTIVITY 293

The net rate of transport per unit area, which we identify with the heat current H,
is
H = _! niic• JdT (IQ-24)
3 • dy.
and by comparison with Eq. (IQ-23) we see that t he thermal conductivity lis

l = ! niic•/
3 •
= !3 iic:
q
(IQ-25)

Thus the thermal conductivity, like the viscosity, should be independen t of

density. This is also in good agreement with experiments down to pressures so low
that the mean free path becomes of the same order of magnitude as the dimen·
sions of the container.
The ratio of thermal conductivity to viscosity is

and
lM =I, (IQ-26)
'f/C•

where M is the molecular weight of the gas. Therefore the theory predicts that for
all gases this combination of experimental properties should equal unity. Some
figures are given h Table IQ-2 for comparison. The ratio does have the right o rder
of magnitude, but we see again that the hard-sphere model for m olecules is in·
adequate.

Table 10-1 Values or the the rmal conductivity A, molecular weight M, viscosity ~. and
specific hear capacity c. or a number or gases

A(0°C) M •r<o•C) c. AM
Gas (J m- 1 s-1 K- 1) (kg kilomole- 1) (N s m-•J (J kilomole-1 K- 1) -~.
He O. t41 4.003 t8.6 X to-• 12.S X to> 2.43
Ne .0464 20. t8 29.7 t2.7 2.48
A .163 39.9S 21.3 t2.S 2.45
H, . t68 2.0t6 8.4t 20.t 2.06
N, .24t 28.02 t 6.6 20.9 I 1.95
o, .24S 32.00 t 9.2 21.0 1.94
co, . t4S 44.01 t3.7 28.8 1.62
NH1 .2t8 t7.03 9.2 27.6 1.46
CH, .lOS 16.03 t0.3 27.4 1.73
Air .24 1 29. 17.2 20.9 1.94
 294 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA 10-6

10-8 DIFFUSION

The vessel in Fig. 1()...9 is initially divided by a partition, on opposite sides of which
are two different gases A and Bat the same temperature a nd pressure, so that the
number of molecules per unit volume is the same on both sides. If the partition
is removed, there is no large scale motion of the gas in either direction , but after a
sufficiently long time has elapsed, one finds that both gases are uniformly distrib-
uted throughout the entire volume. This phenomenon, as a result of which each
gas gradually permeates the other, is called diffusion. It is not restricted to gases
but occurs in liquid and solids as well. Diffusion is a consequence of random
molecular motion and occurs whenever there is a concentration gradient of any
molecular species, that is, when the number of particles of one kind per unit
volume on one side of a surface differs from that on the other side. The phe-
nomenon fan be described as a transport of malltr, (that is, of molecules) across
a surface.

• • • • •••
••••••••
••·.·.·.s.·.·.·
••••••••••••
•·.•.•.•.·.·.·

Fig. ID-9 A vessel con-

taining two different
gases separated by a par-
tit ion.

The phenomenon of diffusion may be complicated by the fact that when more
than one type of molecule is present the rates of diffusion of one into the other
are not the same. We can simplify the problem and still bring out the essential
ideas by considering the diffusion of molecules of a single species into others of the
same species, known as ulfdiffusion.
If all of the molecules of a system were exactly alike, any calculation of self-
diffusion among them would be of academic interest only, since the re would be no
experimental method by which the diffusing molecule& could be distinguished
from the others. However, molecules that are isotopes of the same element, or
molecules whose nuclei have been made radioactive, differ only in their nuclear
structure and are essentially identical as far as collision cross sections are concerned.
 DIFFUSION 295

(Their mean kinetic energies will differ slightly because of differences in mass.)
It is thus possible to "tag" certain molecules so that they can be distinguished
from others, and yet treat the problem as if the molecules were all alike.
Consider an imagined horizontal surface S·S within the vessel of Fig. 10-9,
at some stage of the diffusion process. The vessel contains a mixture of tagged and
untagged molecules, the total number of molecules per unit volume being the same
at all points so that the pressure is uniform. We ossume the temperature to be
uniform also. Let n* represent the number of tagged molecules per unit yolume
at any poinr. We shall assume that n* is a funclion of y only, where the ylaxis is
normal to the surface S·S. If tfn*fdy is positive, the downward fl ux of tagged
molecules across the surface is then greater than the upward flux. If r represents
the net flux of tagged molecules across the surface, per unit time and per unit area,
the cotfficimt ofsrlfdi.ffusion D is defined by the equation
r = - D dn* • (10-27)
dy
The negative sign is included since if dn*/dy is positive, the net flux r is down-
ward and is negative.
In the MKS system, the unit of r is I molecule per square meter per second and
the unit of the concentration gradient dn*/dy is I molecule per cubic meter, per
meter. The unit of the diffusion coefficient Dis therefore I molecule per square meter
per second, per molecule per cubic meter, per meter, which reduces to I m' s-1•
We assume as before that each molecule makes its last collision before crossing
at a perpendicular distance 21/3 away from the surface. If n6 is the number of
tagged molecules per unit volume at the surface S·S, the number per unit volume at
a distance 21/3 below the surface is •
n• = n:- ~/!!!!..._
3 dy
In the expression previouslyderived for the flux «1>, we must replace n by n•, and
the upward flux rr is then *)
rr =!4 v(n.r- 3~ 1~
tly
•

In the same way, the downward flux is

r! = !
4
v(n: +~, tin*).
3 dy
The net flux r is the difference between these, so
r- - ! i!/ dn* . ( 10-28)
3 dy
Comparison with Eq. (10-24) shows that
D=!vl=!.!., (10- 29)
3 3 no
where n is the total number of molecules per unit volume.
 21f INTERMOLECULAR FORCES. TRANSPORT PHENOMENA 1()..7

The phenomenon of diffusion through fine capillary pores in a ceramic material

is one of the methods used to separate the isotopes u= and U 131• Naturally
occurring uranium is converted to the hexafluoride UF,, a gas, and the mixture
of isotopes flows by diffusion through a porous barrier. The phenomenon is more
complicated than the simple case descri bed above because the free path is no
longer small compared with the dimensions of the capillaries, and collisions with
the walls become an important factor. However, we can see qualitatively that
because of the slightly smaller mass of U 135 compared with U 131, the mean speed jj
of the hexafluoride molecules containing U'"
will be slightly greater than for the
others. The diffusion coefficient is slightly greater also, so that this component is
slightly enriched in the gas that has diffused through the pores.
The operation of a nuclear reactor is also dependent on the phenomenon
of diffusion. The neutrons in a reactor behave like a gas that is continuously being
generated throughout the reacto r by fission processes and which diffuses through
the reactor and eventually escapes from the surface. In order that the reactor may
operate successfully, conditions must be such that the rate of generation of neu-
trons is at least as great as the loss by diffusion, p lus the losses due to collisions
in which the neutrons are absorbed.

10-7 SUMMARY
Let us compare the three results obtained in the preceding sections. We can wri te
Eqs. (10-20), (10-24), and (10-28) as
G
-
(Ij nu_1)"'dJ,
d(mu)

H- -(~ nut)d(c:T)
3 dy •
r ... - (!3 nvt) d(n*/n).
dy
T he last equation is obtained by multiplying numerator and denominator of
Eq. (I 0-26) by n.
The product (mu) in the fi rs t equation is thejloll' momentum of a gas molecule,
the product (c:TJ in the second is the kinetic tnugy of a molecule, a nd the ratio
(n"fn) in the third is the conctntration of tagged molecules.
The corresponding expressions for the coefficients of viscosity, of thermal
conductivity, and of self·diffusion, are
I - I iim
1J = 3 nmul = )-;•

l - ! nc•Vt :c: ! Vc:

3 ' 3 a
v~ vt =!i..
1
3 3 no
 PROBLEMS 297

PROBLEMS

16-1 How arc the assumptions of kine tic theory given in Section 9-2 changed in the
developm<nt of the Hirn and the van der Waals equations of state?
16-l The critical temperature of C01 is 3 1.1°C and the critical pressure is 73 atm.
Assume that C01 obeys the van der Waals equation. (a) Show that the critical density
of C01 is 0.34 g em- •. (b) Show that the diameter of a C02 moloculo is 3.2 >< m. JO-••
10-3 Using the data of the previous problem, (a) find the m icroscopic collision cross
section for a C02 molecule. (b) If on< kilomole o f C02 occupies 10 m•, find the mean
free path of theco,molecules. (c) If the mean speed of tho C01 molecules is SOO m s-•,
compute the average number of collisions made per molecule in one second.
1o-4 Find the pressure dopendence at constant temperature of the mean free path and
the collision frequency.
16-5 A beam of molecules of radius 2 >< tO-•• m strikes a gas composed of molecules
whose radii arc 3 >< to-•• m. Thore are 10" gas moleculos per m•. ~termine (a) the
exclusion radius, (b) the microscopic collision cross section, (c) the macroscopic collision
cross section, (d) the fraction of the beam scattered per unit distance it travels in the gas,
(e) the fraction of molecules left in the beam after it tra vels to-• m in the gas, (f) the
distance the beam travels in the gas before half of the molecules are scauered o ut, (g)
the mean free path of the beam in the gas.
1()-6 A group of oxygen molecules start their free paths at the same instant. The pressure
is such that the mean free path is 3 em. After how long a time will half of the group still
remain unscallered. Assume all particlos have a speed equal to the rms speed. The tern·
perature is 300 K.
16-7 Dowling pins with an etfective diameter of tOem are placod ra ndomly on a bowling
green with an average density of 10 pins per square meter. A large number of IO·cm
diameter bowling balls are bowled at the pins. (a) What is the ratio of the mean free path
of the bowling ball to the average distance between pins? (b) What fraction o f t he bowling
balls will travel at loast 3 meters without striking a pin?
10-8 The moan free path in a cortain gas isS em. Consider 10,000 mean fret paths. How
many are longe r tha n (a) S em? (b) 10 em? (c) 20 em 7 (d) How many arc longer than
3 em but shorter than S em? (e) How many a re between 4.S and S.S em long? (f) How
many are betwee n 4.9 and S. l em lo ng? (g) How many ore exactly S em long?
16-9 A large number of throws arc made with a single die. (a) Wh at is the average
number of throws between the appearances of a six? At any stage or the process, V!hat is
the ave:"ttge number of throws (b) before the noxt appearance of a six, (c) since the last
appearanco of a six? (d) How do you answer the question raised in Section I G-3; that is,
why is the mean free path I and not 2/?
10- 10 The mean free path of a helium atom in helium gas at standard conditions is
20 >< lo-' m. What isthe radius of a holium atom?
tO- ll A beam of olectrons is projected from an electron gun into a gas at a pressure P,
and the number remaining in the beam at a d istance x from the gun is determined by
allowing the beam to strike a collecting plate and measuring the current to the plate. The
I
 218 INTERMOLECULAR FORCES. TRANSPORT PHENOMENA

electron current emitted by the gun is 100 JAA, and the current to the collector when x •
10 em and P • 100 N nr• (lbout I Torr) is 37 ~tA. (a) What is the electron mean free
path? (b) What current would be collected if the pressure were reduced to SON m-•?
10-12 A singly cha'l:ed oxygen molecule staru a free path in a direction at right angles
to an electric field of intensity 10' V m-•. The pressure is one atmosphere and the tem-
perature 300 K. (a) Compute the distanoc moved in the direct ion of the field in a time
equal to that required to traverse one mean free path. (b) What is the ratio or the mean
free path tcj this distance? (c) What is the average speed in the d irection or the field? (d)
What is the ratio of the rms speed to this speed? (e) What is the ratio of the energy of
thermal agitation to the energy gained from the field in one mean free path 1
10-13 The resistance ofl m ofO.Ol em diameter copper wire is measured to be 3 ll. The
density of copper is 8.9 x 10' kg m-• and its atomic weight is 64. (a) Determine the mean
free time ~ between collisions of the electrons with the copper ion cores. (b) Determine
the mean free path of the electrons assuming that ii for an eleclron is given by (8/<"Tf•mr)"'·
How many atomic distances is this, assuming copper is cubic? (c) Determine !he ratio
of the diameter of the copper ion cores to the atomic d istance. (Parts (b) and (c) do not
give correct answers because electron speeds are approximately I o> times as large as those
given by (8kT/•m)1' '· Section 1~.) (d) Determine the average length of time it takes an
electron to move the length or the wire when the current through the wire is 0.333 A .
10-14 Satellites travel in a region where the mean free path of the particles in the acmos-
phere is much greater than the characceristic size of che body. Show that the force per
unil area on che sacellite due to chis rarefied gas is 4nnru1/3, where n is !he number density
of particles in the atmosphere, rn is their mass, and u is the speed or che satellice. [Hint:
Since the sa tell ice speed is much greater than the speed or sound, assume that the sa!ellite
is moving through a stationary cloud of particles.]
10-15 Calculate the coefficien! or friction o r a disc gliding on an air table wich a speed
of l m s-•. The diameter of !he disc is 0.1 m and ics mass is 0.3 kg. Assume !hac il glides
Ill' m above the table. The diameter of a nitrogen molecule is about 4 x 10-•o m.
10-16 The viscosity or carbon dioxide over a range or tempera cures is given in the table
below. (a) Compu!e !he ratio ~/Vr at each lempera!ure an d (b) determine the diameler
of the CO, molecule. (c) Compare that diamecer wich the diamecer o f A and No taken
from Fig. 10-8.

10-17 (a} Derive an expression for chc temperature dependence of !he chermal con-
ductivity of an ideal gas. (b) Calcula!c the thermal conductivicy or helium (considered as
an ideal gas) at 300 K.
10-18 (a) From the d31a in Table 10-2 de!ermine the self-diffusion coefficient of helium
a !Standard conditions in two ways. (b) How does the self-diffusion coefficient depend upon
pressure a! constant lemperacure, upon cemperature at constant pressure, and upon the
mass or the diffusing particle.
 PROBLEMS 2H

IG-19 A tube 2m long and JO-< m1 in cross section contains CO, at atmospheric pressure
and at a temperature of 300 K. The carbon atoms in one-half of the C02 molecules ar e
the radioactive isotope C". At time 1 - 0, all of the molecules at the extreme left end of
the tube con ta in radioact ive carbon, and the number o f such molecules per un it volun1e
decreases uniforn1ly to zero at the other end of Ihe tube. (a) What is Ihe initial concentra-
tion gradient of radioactive molecules? (b) Initially, how many radioactive molecules
per second cross a cross section at the midpoint of the tube from left 10 righ t? (c) How
n1any cross fron1 right to left? (d) What is the initial net rate of diffusion of radioactive
molecules across the cross section, in n1olecules per second and micrograms per second ?
IG-20 Given that the density of air is 1.29 kg m-•, v - 460 m s-•, and I - 6.4 x l o-1 m
at standard conditions, determine the coefficients of viscosity, (b) diffusion, and (c) thermal
conductivity. Assume that air is a diatomic ideal gas.
10- 21 There is a small uniform pressure gradient in an ideal gas at constant temperature
so that there is a mass flow in the direction of the gradient. Using the mean free path
approach show that the rate of flow of mass in the direction of the pressure 4radient per
unit area and per unit pressure gradient is miil/3k T.
 11
Statistical thermodynamics
11-1 I NTRODUCTION

11- 2 ENERGY STATES AND ENERGY L EVELS

1 1-3 MACROSTATES AND MICROSTATES

11-4 THERMODYNAMIC PROBABIUTY

1 1-6 THE BOSE· EINSTEIN STATISTICS

11-8 THE FERMI· DIRAC STATISTICS

1 1-7 THE MAXWELL· BOLTZMANN STATI STICS

11-8 THE STATISTICAL INTERPRETATION OF ENTROPY

11-9 THE BOSE· EINSTEIN DISTRIBUTION FUNCTION

11-10 THE FEAMI· DIRAC DISTRIBUTION FUNCTION

11-11 THE CLASSICAL DISTRIBUTION FUNCTION

11-12 COMPARISON OF DISTRIBUTION FUNCTIONS FOR

INDISTINGUISHABLE. PARTICLES

11-13 THE MAXWELL· BOLTZMANN DISTRIBUTION FUNCTION

11- 14 THE PARTITION FUNCTION

11- 15 THERMODYNAMIC PROPERTIES Of A SYSTEM

 302 STATISTICAL THERMOOYNAMICS 11-2

11-1 I N~RODUCTION
The metliods of statistical thermodynamics were first developed during the latter
part of the last century,largely by Boltzmann in Germany and Gibbs in the United
States. With the advent of the quantum theory in the early years of the present
century, Bose• and Einsteint, and Fermi! and Dirac§ introduced certain modifica-
tions of Boltzmann's original ideas and succeeded in clearing up some of the
unsatisfactory features of the Boltzmann statistics.
The statistical approach has a close connection with both thermodynamics
a nd kinetic theory. For those systems of particles in which the energy of the
particles can be determined, one can derive by statistical means the equation of
state of a substance and its energy equation. Statistical thermodynamics provides
an additional interpretation of the concept of entropy.
Statistical thermodynamics (also called statistical mechanics), unlike kinetic
theory, does not concern itself with detailed considerations of such things as
collisions of molecules with one another o r with a surface. Instead, it ta kes ad-
vantage of the fact that molecules are very numerous and a!N'rage properties of a
la rge number of molecules can be calculated even in the absence of any information
about specific molecules. Thus an actuary for an insurance company can predict
with high precision the average life expectancy of all persons born in the United
States in a given year, without knowing the state of health of any one of them.
Statistical methods can be applied not. only to molecules but to photons, to
elastic waves in a solid, and to the more abstract entities of quantum mechanics
called wave functions. We shall use the neutral term "particle" to designate any
of these.

11-2 ENERGY STATES AND ENERGY LEVELS

The principles of classical mechanics, or Newtonian mechanics, describe correctly
the behavior of matter in bulk, or of marroscopic systems. On a molecular or
microscopic scale, classical mechanics does not apply and must be replaced by
quantum mechanirs. The principles of quantum mechanics lead to the result that
the energy of a particle, not acted on by some conservative force field such as a
gravitational, electric, or magnetic field , ca nnot toke on a ny arbitra ry value, or
cannot change in a contimlous manne r. Rather, the particle can exist only in some
one of a number of states having specified energies. The energy is said to be
quantiud.
A knowledge of quantum mecha nics will not be assumed in this book. We
shall try to make pla usible some of its predictions; others will simply be stated and

• Satyendranalh Bose, India n physicist (1894-1974).

t Alben Einstein, German physicist, (1879-1955).
~Enrico Fermi, llalian physicisc (1901- 1954).
§Paul A. M. Dirac, English physicist (1902-
 11-2 ENERGY STATES AND ENERGY LEVELS 303

~-------L------~

Fig. 11- 1 Three of the possible station-

ary waves in a stretched string fixed at
both ends.

the reade r will have to take them on faith or refer to texts devoted to this1subject.
In any event, as far as the methods of statistics are concerned, it is enbugh to
know that quantized energy states exist.
In quantum mechanics, also known as wave mechanics, the general method
of attacking a problem is to set up and (hopefully) solve an equation known as
Schrodinger's• equation. In many problems, this equation is exactly analogous
to the wave equation describing the propagation of transve rse waves in a stretched
string, fixed at both ends. As is well known , the string can vibrate in a steady state
in any one of a number of stationary waves, three of which are shown in Fig. I 1- 1.
That is, there may be a node Nat each end and an antinode A at the center, or
there may be a node at the center as well as at the ends, with anti nodes midway
between the nodes, and so on. The important result is that there is always an
integral numliu of antinodes in the steady-state modes of vibration ; o ne antinode
in the upper diagram, two in the next, and so on . The distance between nodes (or
antinodes) is one-half a wavelength, so if L is the length of the string, the wave-
lengths l of the possible stationary waves are

)., = 2L, ', = 1 2L, et c.;

A
3
or in general
A1 =.!. 2L,
n,
• Erwin Schrodinger, Austrian physicist (1887-1961).
 304 STATISTICAL THERMODYNAMICS 11- 2

where n1 is an integer equal to the number of anti nodes and can have some one of
the values
n1 = 1, 2, 3, ....
A stationary wave is equivalent to two traveling waves propagating in opposite
directions, the waves being reftected and re-reftected at the ends of the string.
This is analogous to the motion of a particle moving freely back and forth along
astraight line and making elastic collisions at two points separated by the distance
L. According to quantum mechanics, a s tationary Schrtxlinger wave is in fact
completely equivalent to such a particle, and t he wavelength .<of the stationary
wave is related to the momentum p of the particle through the rela tion

h
p=:i· (ll-1)

where his a universal constant called Planck's constant. ln the MKS system,

h = 6.6262 X 10..... ] s.
The momentum of the particle is therefore permitted to have only some one o f
the set or val ues
h
p, .. n1 2L·· (11-2)

If a particle is free to move in any direction within a cubical box of side length
L whose sides a re parallel to the .x, y, z axes of a r<etangular coordinate system,
the .x,y, and z components of its momentum are permitted to have o nly the values

where n,, n,, and n, are integers called quantum numbers, each of which can have
some one pf the values I, 2, 3, etc. Each set of quantum numbers therefore corre·
sponds to Ia certain direction of the momentum. Then if p, is the resultant momen-
tum corresponding to some set of quantum numbers n,, n,, n.,

P'J = p'z + p'v + p'' = (n'~ ~•

+ n•,. + n1•) 4LI'
or, if we let (n! + n! + n!) = n:,
I I hi
p, = "'4IJ'
The kinetic energy< of a particle of mass m, speed v, and momentum p - mv
is

<= l
2
mv' =
2m
L.
 11-2 ENERGY STATES AND ENERGY LEVELS 305

The energy <1 corresponding to the momentum PJ is therefore

p" h' ,
~~ = --L
2m
~ n~--.
8mL'
(11-3)

The values of n., n,, and n, are said to define the state of a particle, and the
energies corresponding to the different possible values of are the possible energy n:
levels. The energy levels depend only on the values of n~ and not on the individual
values of n,, n,, and n,. In other words, the energy depends only on the magnitude
of the momentum p1 and not on its direction, just as in classical mechanics. In
general, a number of different states {corresponding to d ifferent di rections of the
momentum) will have the same energy. The energy level is then said to be de-
generate, and we shall use the symbol g1 to designate the degeneracy of level j,
that is, the number of states having the same energy <1.
The volume Vof a cubical box of side length L equals L' , soL' = V'13 ; and
Eq. ( 11- 3) can be written, for a free particle in a cubical box,

• J -- n'J .!!: v-"·• (11-4)

8m
The same result applies to a container of any shape whose d imensions 3re large
compared with the wavelength of the Schrlldinger waves. The energy of the jth
level therefore depends on the quantum number n 1 and on the volume V. If the
volume is decreased, the value of a given <j increases.
As an example, consider a 1-liter volume of helium gas. When numerical values
of h, m, and V are inserted in Eq. (11-4), we find that
h'
Bm v-II• ""8 X 10""<0 J ""s X JO-tl eV.

We have shown that at room temperature the mean kinetic energy of a gas molecule
is about l/40eVor 2.S x lo-• ev. Hence for a molecule with this kinetic energy,
t 2.5 X JO-I II
n1 "" S X J0-2I <:o< S X 10 ,
n, "" 2.2 x 101•
Thus for the vast majority of the molecules of a gas at ordinary temperatures, the
quantum numbers n1 are very large indeed.
The lowest energy level (j = I) is that for which n, = n, = n, = I. Then
n: = 3 and
(l:Z::-
3h' v-"• •
8m
T here is only o ne state (one set of quantum num bers n., n,, nJ having this energy.
The lowestlevel is therefore nondegenerate and g, = I. The x,y, and z components
of the corresponding momentum p 1 are all equal, and each equals hf2L.
 STATISTICAL THERMOOYNAMICS 11-2

In the next level (j = 2) we might have any one of the following states:

n. n, n,

2 l l
l 2 l
l l 2

Thus in the first o f these states, for example, the components of momentum are
h h
p,, = 2L' p,, = 2L.

In each state, n: - (n! + n! + n!) = 6, and in this level,

~. :., 611' v-"•.

Bm
Since three different states have the same energy, the degeneracy g, = 3.
The preceding discussion of the energy levels and degeneracies of a free particle
in a box is only one example of energy q uantization. Other constraints also leading
to energy quantization will be discussed later.
Figure ll-2 represents in a schematic way the concepts of energy states, energy
levels, and the degene racy of a level. The energy ltvtls can be thought of as a set
of shelves at different elevations, while the energy slalts correspond to a· set of
boxes on each shelf. The degeneracy g1 o f level j is the number of boxes on the
corresponding shelf. If a number of marbles are distributed among the various
boxes, the number in any one box is the number in a particular slalt. Those marbles

(I ) (2) (J) (4) (S)

.. u u LJ u ~ f1• • 1.N. • 2

.. LJ u LJ ~ fhz 4,NJ•)

.. l.d LJ GJ g , • l. NJ • 4

~~ • I {Nondcgtnerate~ N, = S
'•

Fig. 11-2 A schematic representalion of a set of

energy levels •1, their degeneraciesg1 and their occupa-
tion numbers N1•
 11- 3 MACROSTATES ANO MICROSTATES 307

i,n the boxes on any one shelf are in different states, but all have the same energy.
T he total number of marbles in the boxes at any level j is called the occupation
number N1 of that level.
Evidently, the sum of the occup ation numbers N1 over all levels equals the
total n umber of particles N:
~ N, ~ N. (11-5)
'
A lso, since the particles in those states included in any level} all have the same
energy.,. the total energy of the particles in level} is <1N1 , and the total energy E
of the system is
~ < 1 N 1 =E. ( 11-6)
'
If the system is in an external conservative force field such as a gravitational,
electric, or magnetic field, the total energy E will consist in part of the potential
energy£" of the system. If the potential energy is zero, the total energy E is then
the internal energy U and
~ <1 N 1 = U. (11-7)
'
11-3 MACROST ATES AND M I CROSTATES
A number N o f identical entities is called an assembly. The entities may be single
particles, or they may themselves be identical assemblies of particles, in which case
one has an assembly of assemblies, or an ensemble. We shall for the most part
consider only assemblies of single particles, and shall refe r to them as an assembly
or simply as a system.
If the distribution of the particles of the system among its enerll>l states is
k nown, the macroscopic properties of the system can be determined. Thus a
central problem of statistical mechanics is to determine the possible distributions of
particles among energy levels and energy states.
The description of a single·particle assembly depends on whether the particles
of which it consists are distinguishable or indistinguishable. Suppose the assembly
is a sample o f gas and the individual molecules are the particles. Since there is no
way in which the molecules can be labeled, the particles are indistinguishable.
On the other hand, if the assembly is a crystal, the molecules can be labeled in
accord with the positions they occupy in the crystal lattice and can be considered
disti nguishable.
Whether the particles are distinguishable or not, a specification of the number
of particles N1 in each energy level is said to define a macros/ate of the
assembly. For example, the macrostate of the assem bly in Fig. 11-2 is specified by
the set of occupation numbers N1 = S, N, - 4, Ns - 3, N, - 2.
If the particles are indistinguishable, a specification of the total number of
particles in each energy state is said 10 define a microstate of the assembly. Thus if
 308 STATISTICAL THERMODYNAMICS 11-3

the energy states in each level in Fig. 11- 2 are numbered {I), {2), (3), etc., up to the
number of states g1 in the level; and if the particles are indistinguishable, the micro-
state of the assembly is specified by saying that in level 4 there is one particle in
each of the states {3) and (5), and there are no particles in states (I), {2), and (4);
in level3 there is one particle-in states (I), (3), and {4), and no panicle in state {2) ;
in level2there are two particles in state (I) and one particle in each of states (2)
and {3); and in level I there are live particles in the on'ly state in this level.
If one or both of the particles in level 4 were in states other than (3) and (5),
the microstate would be different, but the macrostate would be unchanged since
we would still haveN, = 2. Evidently, many different microstates will correspond
to the same macrostate.
If the particles are distinguishabl~. a specification of the energy state of ~ach
partie/~ is said to define a microstate of the assembly. That is, we must specify not
only how ma11y particles are in each state, but which particles they are. Thus suppose
that the particles in Fig. I 1- 2 are distinguishable and are lettered a, b, c, etc., and
that in level 4 particle a is in state {3) and particle b is in state (5); in level 3,
particle cis in state (I) and particles dand e are in states (3) and {4) respectively, and
so on. The preceding specification, including all levels, describes the microstate
of the assembly. In contrast to an assembly of indistinguishable particles, in which
the microstate would be the same no matter • ·hich particles occupied states (3)
and (5) in level 4, the microstate is now considered different if particles a and b
are interchanged between these states. Also, the microstate would be different if,
say, particles c and din level3 were interchanged with a and bin level 4. In each
such inter$ange we have a different specification of the energy states of the par-
tides and hence a different microstate; although the macrostate does not change
because the occupation numbers of the levels are the same.
If there is more than one particle in a given energy state, an interchange of the
order in which the letters designating the particles is written is not considered to
change the microstate. Thus suppose the two particles in state {I) of level 2 are
lettered p and q. The microstate is considered the same if the letters are written
in the order pq or qp.
The number of microstates that are considered different, for a given set of
occupation numbers, is evidently much greater if the particles are distinguishable
than if they are indistinguishable.

The possible macrostates and microstates of an assembly of particles is analogous

to a table of ages of groups of individuals. As an example Jet us take the number of
children in each grade of an elementary school having a total enrollment of 368
children.
11-3 MACROSTATES AND MICROSTATEk 309

T he grades corre5pond to the energy levels of the sys1em and the specificalion of the
number of children in each gr.~de defines the macrosta te of the system. A differen t
macrostale with the same total number of children would be

L~-~-~-i:_:_n~l___: __~__ ~_· 6_: ~ ~ L__~_--JI·

5
s_7__ __ __7J__ __6_:__
The change in distribution may have several macroscopic consequences: needs for
differen t numbers of teachers, different equipment, different numbers of textbooks,
etc.
The gra des could be further subdivided into clasSC5, that is, in the first macro-
state described there may be 3 first grade classes and 2 second grade classes. These
classes would correspond to the degenerate energy states of each level. T here would
be 3 degenerate states in level I , etc.
If the children were considered as indistinguishable particles (a bad pedagogic
practice), then a microstate of the system would be

A different micros tate of the same macros tate of the system would be

Ahhough the number of children in each <lass was changed, the number o f children
in each grodt remained constant.
However, the distribution

would correspond ro a different macrostate since the number of children in each

grade was changed, even though the total number of children in the school remained
consta nt.
When the children are considered distinguishable particiC5, the microsta te is
different, if Evelyn is in I (a) and Mildred is in l(b), or vice versa, or if both are in l(b).
However, in the last case the microstate is the same if Mild red's name appears on the
class list before Evelyn's or after it.
 310 STATISTICAL THERMODYNAMICS 11-4

11-4 TH ERMODYNAMIC PRDBABIUTY

In the preceding section, no restriction was imposed on the possible ways in which
the particles of an assembly might be distributed among the energy states. In an
isolated, closed system, however, the energy E and the total number of particles
N are both constant. Hence the only possible microstates of such a system are
those that satisfy these conditions.
As time goes on, interactions between the particles of an isolated, closed
system will result in changes in the numbers of particles occupying the energy
states, and, if the particles are distinguishable, will result in changes in the energy
state of each particle. These interactions might be collisions of the molecules of a
gas between themselves or with the walls of the container, or an energy inter-
change between the oscillating molecules of a crystal. Every such interchange
results in a change in the microstate of the assembly, but every possible microstate
must satisfy the conditions of constant Nand E.
The fundamental postulate of statistical thermodynamics is that all possible
microstates of an isolaud assembly art equally probable. The postulate can be
interpreted in two different ways. Consider a time interval/ that is long enough so
that each possible microstate of an isolated, closed system occurs a large number
of times. Let At be the total time during which the system is in some one of its
possible microstates. The postulate then asserts that tht timt inttrval At is tM samt
for all microstatu.
Alternatively; one can consider a very large number.#' of replicas of a given
assembly (an ensemble). At any instant, let A.#' be the number of replicas which
are in some one of the possible microstates. The postulate then asserts that tht
number A.,V is the same for all microstates. The postulate does not seem to be
derivable from any more fundamental principle, and of course it cannot be verified
by experiment. Its justification lies in the correctness of the conclusions drawn
from it.
In terms of the example of the previous section, if all microstates were equally
probable and the population of the school were limited to exactly 368 children, over
many, many years each distribution of children among classes would occur as oflen
as any other. Alrematively, if in a given year one looked at many elementary schools
having a population of 368 children, each distribution of children among classes
would occur with the same frequency. In each case, the examples given in the previous
section would occur the same number of times.
The number of equally probable ll)icrostates that correspond to a given
macrostate k is called the thtrmodynamic probability rr. of the macrostate. (The
symbo l rr comes from the German word for probability, Walrrsclrein/ichktit.
Other symbols are often used, and the quantity is also known as the statistical
count.) For most macrostates of an assembly of a large number of particles, the
thermodynamic probability is a very large number. The total number n of possible
microstatn of an assembly, or the thermodynamic probability of the asstmbly,
 11-4 THERMODYNAMIC PROBABILITY 311

equals the sum over all macrostates of the thermodynamic probability of each
macrostate:

The principles of quantum mechanics lead to expressions for the possible

different ways in Which the particles may be distributed among the energy states of
a single assembly at one instant of time. In other words, quantum mechanics
determines the microstate at each instant for a single assembly or of each of the
large number of replicas of an assembly at one instant. The calculation of ir• for
three different cases is carried out in Sections 11-5, 11 -6, and 11- 7.
The observable properties of a macroscopic system depend on the time
average values of its microscopic properties. Thus the pressure of a gas depends
on the time average value of the rate of transport of momentum across an area.
By the fundamental postulate, the observable properties of a macroscopic system
will also depend upon the average value of the microscopic propenies of a large
number of replicas of an assembly taken at one instant.
Thus the primary goal of a statistical theory is to derive an expression for the
average number of particles IV1 in each of the permitted energy levels j of the
assembly. The expression to be derived is called the average occupation number of
the levelj.
Let N1• be the occupation number of level j in macrostate k. The group
average value of the occupation number of level j, IV!, is found by multiplying N1•
by the number of replicas in macros tate k, summing over all macrostates and
dividing by the total number of replicas, f . The total number of replicas of a given
assembly that are in macrostate k equals the product of the number of replicas
ll.IV that are in some microstate and the number of microstates -ur. included in
the macrostate. Therefore

However.
..v- I ir.Af,
and since A.A·· is 1he same for all m:crostates, we can cancel it from the numJrator
and denominator. The group average is
I N,.-tr. I
IV;,.. t I -;r. =!it N,.ir•. ( 11- 8)
•
Similarly, we can calculate the time average of the occupation number of level
j. IV/. As explained above, the postulate that all microstates are equally probable
means that over a sufficiently long period of time 1, each microstate exists fo r the
same time interval At. The total time the assembly is found in macrostate k is
 31Z STATISTICA~ THERM ODYNAMICS 11-5

then the product of the time interval tlt and the number rr. of microstates in
macrostate k. The sum of these products over all macrostates equals the total
timet :
I - I il'".llt .
•
The tim~ a~rag~ value of the occupation number of level j. Nl. is found by
multiplying the occupation number N1• of level j in macroslate k by the time
rr. tlt that the assembly spends in macrostate k, summing these products over all
macrosta tes, and dividing by the total time 1. The time average is therefore

1
I N,."'r. t>t
n: - -IN,.rr.t>t
,• =•I rr. t>t .
•
Since tlt is the same for all microstates, we can cancel it from numerator and
denominator, giving
I N,.rr. '
n:- ·~~v =-IN,.rr•. (11-9)
~"'• n•
•
Comparison ofEqs. (I 1- 8) a nd (I 1- 9) shows that if all microstates are equally
probable, the timt> avuage value of an occupation number is equal to the group
at•nag~. and we can represent either by N1•
The values of the average occupation numbers of the energy levels are cal-
culattd for diffe rent cases in the next three sections. The general expressions for
the N1 , the distribution functions for these cases, a re derived in Sections I 1- 9 to
11-12.

11-5 THE BOSE-EINSTEIN STATISTICS

The thermodynamic probability tr. of a macrostate of an assembly depends o n
the particular statistics obeyed by the assembly. We consider first the statistics
developed by Bose and Einstein, which for bre vity we shall refer to as the B- E
statistics. In the B-E statistics, the particles are considered indistinguishable, a nd
there is no restriction on the number of particles that can occupy a ny energy state.
The energy states, howeve r, are distinguishable. Let the particles be lettered a, b, c,
etc. (Although the particles a re indistinguishable, we assign letters to them tern·
porarily as an aid in explaining how the thermodynamic probability is computed.)
In some one arrangement of the particles in an arbitrary level j. we might have
particles a and bin state (I) of that level, particle c in state (2), no particles in
state (3), particles d, ~.f. in state (4), and so on. This distribution of particles
among states can be represented by the following mixed sequence of numbers and
letters:
[(l)ab) [(2)c) [(3)) [(4)dif] · · · (I 1- 10)
 11-5 THE BOSE·EINSTEIN STATISTICS 313

where in each bracketed group the letters following a number designate the particles
in the state corresponding to the number.
If the numbers and letters are arranged in all possible sequences, uch sef!uence
will represent a possible distribution of particles among states, provided ~he se-
quence begins with a number. There are therefore g, ways in which the sequences
can begin, one for each of the g, states, and in each of these sequences the remaining
(g1 + N1 - I} numbers and letters can be arranged in a ny order.
The number cif different sequences in which N distinguishable objects can be
arranged is N! (N factorial}. There are N choices for the first term in a sequence.
For each of these there are (N- I} choices for the second, (N- 2} choices for the
third , and so on down to the last term, for which only one choice remains. The
total number of possible sequences is therefore
N(N - I}(N - 2} · · · 1 = N!
As an example, the three lellers a, b, and c can be arranged in the following
sequences:
abc, otb, bca, hoc, cba, cab.
We see that there are six possible sequences, equal to 3 !.
Using the example of lhe previous section, the number "'r of different sequences
in which the 70 children of the first grade can be lined up is 70!. II is shown in Appendix
C that Stirling·s• approximation for the natural logarithm of the factorial of a large
number xis
Jn x! ... x In x - x.
Hence
In 70! • 701n 70-70 • 24S
log10 701 ~ 245/2.303 = 106
101 = 1o•••.
The number of different possible sequences of the (g, +
N 1 - I} .n umbers
and letters is therefore (g, + N,- I} ! and the total number of possible sequences
of g, numbers and N 1 letters is
g1[(g, + N,- 1}!]. (I 1- 1I}
Although each of these sequences represents a possible distribution of particles
among the energy states, many of them represent the same distribution. For
e xample, one of the possible sequences will be the following:
[(3}] [(l}ab] [(4)def] [(2}c] · · ·.
This is the same distribu tion as {I 1-10}, since the same states contain the same
particles, a nd it differs from {I 1- 10} only in that the bracketed groups appear in a
different sequence. There are g, groups in the sequence, one for eac h state , so
the number of different sequences of groups is g) ! and we must divide {I 1- 1 I} by
g,! to avoid counting the same distri bution mo re than once.

• James Stirling, Sconish mathematician (1696- 1770).

 314 STATISTICAL THERMOOYNAMICS 11-5

Also, since the particles are actually indistinguishable, a different sequence of

lel/ers such as
[(l)ca] [(2~] [(3)) [(4)bdf) · · •
also represents the same distribution as (11- 10) because any given state contains
the same number of particles. The N1 letters can be arranged in sequence in N11
different ways, so ( I 1-1 I) must also be divided by N1 !. Hence the number of
different distributions for the jlh level is

... ,- g [(g g,+!N,I

1 N 1 1 - l)!]

which may be more conveniently written as

(g1 + N1 - 1)1
(11- 12)
WJ :=r (gl- 1}! Nlf '
since
g,l - g,(g, - 1)1.
As a simple example, suppose that an energy level j includes 3 Slates (g1 - 3)
and 2 particles (N, - 2). The possible distributions or the particles among the states
are shown in Fig. 11-3 in which, since the particles are indistinguishable, they are
represented by dots instead or letters. The number or possible distributions, from Eq.
(11-12), is (3 + 2 _ l)l 41
"'I - (3 - 1)!21 ~ 2!21 - 6
•
In agreement with Fig. 11-3.

s.... (1) (2) 131

Fig. 11-3 The pos-

sible distributions of
two indistinguish-
able particles among
three energy states,
with no restriction
on the number of
particles in each
state.
 11-5 THE BOSE- EINSTEIN STATISTICS 311

If a level is nondegenerate, that is, if there is only one state in the level and
g1 ~ I, then there is only one possible way in which the particles in the level can
be arranged, and hence '"' = I. But if g1 = I, Eq. (I 1- 12) becomes
N,!
to 1 = - - = I.
OIN1 1
It follows that we must set 0! - I , whicp may be considered as a convention that
is necessary in order to get the right answer. A further discussion can be found in
Appendix C.
Also, if a level j is unoccupied and N1 - 0.
(g,- I )! I
«1J = (gl - l )t (0)! =
and w 1 = I for tha t level. -~
For each of the possible distributions in a ny level, we may have any one of the
possible distribu tions in each of the other levels, so the toril number of possible
distributions, or the thermodynamic probability tr0 . 8 of a macrostate in the
B-E statistics is the product over all levels of the values of w 1 for each level, or,

"'I' ~ "'I' =II w = II (g, + N, - J)! ( 11- 13)

D· E • I I I (g,- I)! N,! •
where the symbol rr, means that one is to form the product of all terms following
it, for all values of the subscript}. It corresponds to the symbol I 1 for the sum
of a series of terms.
If an assembly includes two levels p and 'I• wilh g. - 3 and N. - 2, as in the
preceding example, and with g.- 2, N.- l , the thermodynamic probability or the
macrostate N, • 2, N. - 1, is
4! 2!
.,-D·B-
2121
"Jili- 6 X 2 - 12,
and there are 12 different ways in which three indistinguishable particles can be
distributed among the energy states or the assembly.
We next calculate the thermodynamic probabilities of those macrostates that
a re accessible to a given system and the average occupation numbers of the per·
mined energy levels. Although all microstates of an isolated, closed syste m are
equally probablt, the only possible microstates are those in whic h the total numbe r
of panicles equals the number N of particles in the system, and in which the total
energy of the particles equals the energy U of the system. As an example, suppose
that we have a system of j ust six particles, that the perm itted energy levels are
equally spaced, and that there are th ree energy states in each level so tha t g1 - 3.
We shall take the reference level of energy as that of the lowest level, ¥> that
<o = 0, ., ~ •· •• = 2•, etc. We also assume that the total energy U of the system
equals 6•.
If the particles are indistinguishable and the system obeys the B-E statistics,
the only possible macrostatts consistent with the conditions N = 6, U = 6t, a re
 31 6 STATISTICAL THERMOOYNAMICS 11-5

In II Fl, N•6
0041 U • 6t

0.088 n - 1sJ2

0 20S ,,• l

0.410

.. G.tlO
.. 160

c:·:: .... .. . .. 2.1)

""·.- 6l Ill Ill 180 90 270 180 100 216 Ill 28

Fig. 11-4 The eleven possible macroslates of an assemblyof 6 particles obeying

Bose-Einstein statislics. The energy levels are equally spaced and have a
degeneracy g1 = 3 in each level. The total energy of the system is U - 6<. The
thermodynamic probability of each macrostate is given at the bouom and the
average occupation number of each level is printed on the right of the diagram.

shown in the columns of Fig. 11-4. !Oach horizon tal row corresponds to an energy
level (t he t hree states in each level are not shown in the figure). T he dots represent
the number of particles in each level. The columns could representeitherthe macro-
states of a single system at different times, o r the mac rostatesof a numbe r of replicas
of t he system at a given insta nt. If we consider t he figure to represent these replicas,
t hen out of a large number..¥ of replicas there would be a number A.¥ in each
macrostate, but since all of these numbers t,.,V would be equal, we can consider
that each macrostate occurs just once.
The diagram can be constructed as foll ows. The macrostate re presented by
the first column is obtained by first placing one particle in level 6, with e nergy 6<.
The remaining five particles must then be placed in the lowest level with energy
zero, so that t he total energy of the system is 6•. Evidently, there can be no particles
in levels higher than the sixth. In t he seco nd column, we place one part icle in
level 5, one particle in level I , and the remaining four particles in the lowest level,
and so o n.
The thermodyna mic p robability. ""~~'• of each macrostate, calc ulated from Eq.
( 11- 13), is given under t he corresponding co lumn. Thus for macrostate k = I,
since g1 = 3 in all levels and all occupation numbers are zero except in level 6,
where N, = I, and in level 0, where N 0 = 5,

(3 +I - I) ! (3 + 5. - I)!
if", = ! I! · ! = 3 X 21 - 63.
2 2 51
That is, the single particle in level 6 could be in any one of three states, and in the
 11-6 THE FERMI-DIRAC STATISTICS 317

lowest level the remaining five particles could be distributed in 21 different ways
among the three states, making a total of 63 different possible arrangements.
The total number of possible microstates of the system, or the thermodynamic
pro ba bility of the system, is
n = I ,-. ...
1532.
•
The average occupation numbers of each level, calculated from Eq. (11-8),
are given at the right of the corresponding level. In level 2, for example, we see
that macrostate 3 includes 135 microstates, i n each of which there is one particle
in level 2. Macrosta te 6 includes 270 microstates in each of which theJe is also
one particle in level 2, and so on. The average occupation number of level 2 is
therefore
1272
n. - n.r
.!. N ....,.. - 1532 o.83.=
I n a ny macrostate k in which level 2 is unoccupied, the corresponding value
of N, is zero and the product N,.'fr . for that level is ze ro. Note that although the
attual occupation number of any level in any macrostate must be a n integer or
zero, the atJtrage occupation nu mber is not necessarily an integer.
The most probable matrostate in Fig. 11-4, that is, the o ne with the largest
number of microstates (270), is the sixth. The occupa tion number of each level
for this macrostate is roughly the same as the average occupatio n number for the
assembly. It can be shown (Appendix D) that when the number of particles in an
assembly is very large, the occupation numbers in the most probable state are very
nearly the same as the average occupation numbers.

11 - 6 T HE FERMI-DIR AC STATISTICS
The statistics developed by Fermi a nd Dirac, which for brevity we call the F-D
statistics, applies to indistinguishable particles that obey the Pauli• exduslon
prindple, acco rding to which there can be no more than one particle in each
permitted energy state. (It is as if e very particle were aware of the occupancy of all
states, a nd could o nly take a state unoccupied by any other particle.) Thus the
a rrangements in the upper th ree rows of Fig. 11-3, in which there are two particles
in each state, wo uld not be permitted in the F·D statistics. Evidently , the number
of particles N1 in a ny level cannot exceed the number of states g1 in that level.
T o calculate the thermodynamic probability of a macrostate, we again tem-
porarily assign numbers to energy states of a level and letters to the particles, and
we rep resent a possible a rrangement of the particles in a level by a mixed sequence
of numbers and letters. A possible arrangement might be the following:
[(J)a] [(2)b] [(3)] [(4)t] [(5)] · · · ( I 1-14)
meaning tha t states (I), (2), (4), .. ·. are occupied with their quota of one pa rticle
each while states (3), (5), .. . are empty. Fo r a given sequence of numbers, we
• Wolfgang Pauli, Austrian physicist (1900-1958).
 318 STATISTICAL THERMODYNAMICS 11-6

first select some arbitrary sequence of lellers. There are g1 possible locations for
the first Jetter, following any one of the g1 numbers. This leaves only (g1 - I)
possible locations for the second Jetter, (g1 - 2) locations for the third, down to
[g1 - (N1 - I )] or (g1 - N1 + I ) locations for the last Jetter. Since for any one
location of any one Jetter we may have any one of the possible locations of each
of the others, the total num ber of ways in which a given sequence of N1 letters can
be assigned to the g1 states is
g,l
g1(g1 - l)(g1 - 2) · · · (g1 - N 1 +
I) = (g, _ N,)J , (11-15)
since
g1 ! = g,(g1 - l)(g1 - 2) · · · (g1 - N1 + l )(g1 - N 1)!.
Because the particles are indistinguishable, a state is occupied regardless of the
particular Jetter that foll ows the num ber representing the state, and since there
are N/ different sequences in which the N1 letrers can be writren, we must divide
Eq . (11- 15) by N1 !. Again, although the sta tes are distinguishable, a differen t
sequence of states does not change the distribution. Therefore we do not need to
consider other sequences of letters a nd for level j,
g,!
«>, = (g 1 - N 1)1 N/
. (11-16)

If a level j includes 3 stales (g 1 - 3) and two particles (N1 ~ 2), then

31 3!
3
WJ - (3 - 2)!2! - I !2! = ·
The possible arrangements are shown in Fig. 11-S, which corresponds to the
lower three rows of Fig. 11- 3, the upper three being excluded.
Finally, since for every arrangement in any one level we may have a ny one of
the possible a rrangements in the other levels, the the rmodynamic probability

Staceffi§(ll
(2) (J)
.
.
Fig. JJ -5 The pos-
sible distributions
of two indistinguish-
able particles among
three energy stales,
with no more than
one particle in each
state.
 11-6 THE FERMI-DIRAC STATISTICS 311

l 11, N • 6
fj/f - 4 • ( 0.12.1 U•6t

0.494 n- n
... • I.I.S IJ •l

I • ·. ) l .ll

0 ••• •••• :~ • • 2$1

- . - T l f " "9 I 11

Fig. 11--6 The five possible macrostalcs or an

assembly or 6 parriclcs obeying Fermi-Dirac
slalisrics. The energy levels arc equally spaced
and have a degeneracy or &1 - 3 each. The
Ioiii energy or rhe syslem is U - 6<. The
lhermodynamic probability or each macroS! ale
is given a1 Ihe bollom, and rhe average occupa-
lion number or each level is prinred on rhe
righr of rhe diagram.

if'" F-o of a macrostate in the F-0 statistics is

if'"F-D - .,-. . . IT "'I - IT gl! (11- 17)
I I (gl- Nl)l Nl!
Figure 11-6 shows the possible macrostates of a system of six particles obeying
the F-D statistics in which, as in Fig. 11 -4, the energy levels are equally spaced
and the degeneracy of each level is g1 - 3. In comparison with Fig. 11- 4, macro-
states I, 2, 3, 5, 10, and II of that figure are excluded because there can be no
more than three particles in each leveL Thus there are only five possible macrostates,
each with energy 6<. The thermodynamic probability of each macrostate, calculated
from Eq. (11- 17), is written underthecorresponding column. Thus in macrostate I,

tr, - (3 - 3!I)! !! (3 - 3!2)! 21 (3 - 3!3)! 31 = 3 X 3 X l = 9.

That is, there are three possible locarions of the single particle in level 4 (in
a ny one of the rhree srares), rhree ways in which the two particles in level I can be
distributed amongrhe rhree slales (as in Fig. 11-S)and only one way in which rhe rhree
parricles in level zero can be dislribured among lhe three slales (one in each slate).
The total number of possible macrostates is
n= I.• , . .
= 73.
The average occupation numbers of each level, calculated from Eq. ( 11-8)
are given at the right of the corresponding leveL These mny be compared with the
occupation numbers in Fig. 11-4.
 320 STATISTICAL THERMODYNAMICS 11-7

Sl&l< (I) (2) (J)

I
""
II

II
""
""
v •
.
6

. .
v 6

vI b

..
VI I b

VII I b

IX b

F ig. 11-7 The pos·

sible arrangements
or two distinguish-
able particlesaandb
among lhree energy
states, with no re·
striction on the
number or particles
per SIBle.

11-7 THE MAXWELL- BOLTZMANN STATISTICS

In the Maxwell-Boltzmann stalislics, which for brevity we call M-B slatislics, the
parlicles of an assembly are considered distinguishable, bul as in the B-E stalistics
t here is no reslriclion on the number of particles thai can occupy the same energy
stale. We consider an assembly of N particles and a macroSiate specified by the
occupation numbers N,. N,, . . . , N1, • • • • The degeneracies of lhe levels are
respectively g,. g,, ... , g1, • • • • Since the particles are distinguishable, two
a rrangements a re considered different if a level contains different particles, even
though the occupation number of the level may be the same. That is, an arrange-
ment in which t he particles in a level are a, b, and cis different from one in which
the particles are a, b, and d or p, q, and r. Consider first any level j, including g1
states and some specified set of N1 particles. T he first particle may be placed in
any one of the g1 states. But since there is no restriction on the number of particles
per state, the second particle can also be placed in any one of the g 1 stales, making
a total of c: possible locations for the first two particles. Since there a re N1
particles in the level, the total number of possible distributions in this level is
(1 1-18)
 11-7 THE MAXWELL-BOLTZMANN STATISTICS 321

For example, if level j includes three s1a1es (K1 - 3) and the two particles a and b
(N1 - 2), the possible arrangemeniS of !he parriclcs arc shown in Fi,. 11- 7, and we
sec !hat there arc nine. An inlcrchange of the letters a and b belwccn different Slatts,
as in arrangcmcnls I V and V, VI and VII, VIII and IX, is considered to give ri~ to a
diffcrcnl microstalc since the particlu a a nd b arc in different states. On the other
hand, a change in the order of the letters within a given state docs not change the
microstate since it leaves the same panicles in the same stale. That is, in arrange-
menr. J, II, and JJJ we could equally well have designated the panicles as ba instead
of ab. Note that if the particles are indistinguishable and are represented by dols
instead of lctlcrs, arra ngements IV and V correspond to the same microstates, as do
arrangcmcniS VI and Vll, and VJII and IX, leaving only six different arrangements as
in Fig. I 1-3. From Eq. (1 1-18), the number of different arra ngements is
in agreement wilh Fig. 11-7. !{'- 3' ~ 9,
Since for a ny d istribution of panicles in o ne leve l we may have any o ne of the
possible distributions in each of t he o ther levels, the toral number of distributions
including all levels, with a specified set of particles in each level, is I
III "'' = II
I
gf•. (11- 19)
But II,w 1 is not equa l to tr. as in t he othe r statistics since a n intercha nge of
particles between ~~~Is (as well as an inlerchange between states in the same level)
wiU also give rise to a different microstate. {If t he particles a re indistinguishable,
an interchange bel ween levels does not result in a diffe rent microstate.) T hus for
example, if ·particle b in Fig. 11-7 were interc hanged with particle c f rom some
o t her level so t hat t he t wo particles in level j we re a and c instead o f a a nd b, we
would have another nine different a rrangements of particles in this level. T he
question then is, out of a total of N particles, in how many different ways can t he
particles be distributed among the e nergy levels, with given numbers of particles
N, N,, N,, etc., in the various levels?
I magine that the N letters representing the particles are written down in a ll
possible sequences. We have shown that there a re N! such sequences. Let the
first N 1 letters in each sequence represent the par ticles in level I, the next N 1
letters those in level 2, and so on. Out of theN! possible sequences, there will be
a number in which t he samt letters appea r in t he fi rst N 1 places, but in a different
order. Whatever the order in which t he letters appea r, the same particles are
assigned to level I, so we must d ivide N! by the number of different sequences in
which the same letters appea r in the first N 1 places, which is N 1 !. In the same way,
we must also di vide by N 2 !, N,!, etc., so that the total number of ways in which N
particles can be distributed amo ng the levels, with N , particles in level I, N,
particles in level 2, and so on, is
N! N!
(ll- 20)
N, !Na!· ·· - fi N,!.
I
The total number of different dist ributions, or the thermodynamic probability
11'",._8 of a macrostate in the M-B statistics, is therefore the product of ( I 1- 19)
 322 STATISTICAL THERMOOVNAMICS 11-7

10 II 1'/1

•I• • 6!---'-1--f--t-~--t-·+--l-+-t--t---iO.OIJ U• 6•
s '
1-+- - -~-

-· 0910

Fig. 11-8 The eleven possible macros1a1es or an assembly or 6 particles obeying

Ma•weii-Bollzmann statistics. The energy levels are equally spaced a nd have a
degeneracy org1 ~ 3 each. The total energy or the system is U - 6<. The therm~
dynamic probability or each macrostate is given at the bouom, and the average
occupation number or each level is printed on the right or the diagram.

and {11-20):
1(/'AI· R - -Nl- IT gl''"' = N I. II-gJ"'. (11- 21)
IT N,! I I Nl!
I
Figure 11-8 shows the possible macrostates of an assembly of 6 particles
obeying the M-B statistics. As in Figs. 11-4 and 11- 6, the energy levels arc pre-
sumed to be equally spaced and the degeneracy of each level is g1 = 3. Although
each panicle could be designated by a leuer, the dots represent only the occupation
numbers N, or the respective levels. The figure is identical with Fig. 11-4 for the
B-E statistics, but it represents a much greater number of microstates because of
the possible interchanges of particles between the states in any level, and between
various levels. The thermodynamic probability of each macrostate, calculated
from Eq. (11- 21), is given under the corresponding column. The values of iJ"•
have been divided by 3'. T hus for mncros ta te k = I, in which only levels zero and
si• a re occupied,
31 31 5
iJ"1 = 6!-- = 18 X 3 ,
S! I!
1r.f3' = 18.
The total number of possible microstates is
Q =! if"• = 1386 X 3' ~ 3.37 X JO'.
•
The average occupation nu mber of each level is given at the right of the corre-
sponding row.
 11-$ THE STATISTICAL INTERPRETATION OF ENTROPY 323

11__. THE STATISTICAL INTERPRETATION OF ENTROPY

In the three preceding sections, the average occupation numbers of the energy
levels of a system were calculated for particles obeying the Bose-Einstein, Fermi·
Dirac, and Maxwcll-lloltzmann statistics. It was stated in Section 11-4 that the
thermodynamic variables of a system were related to t he average occupation
numbers of its energy levels. In this section we derive the connection and begin
by asking what property of a statistical model of a system can be associated with
its entropy.
For two equilibrium states of an op~n PVT system in which the temperature,
pressure, and chemical potential arc t he same but in which the energy, volume,
and number of particles are different, the principles of thermodynamics lead to the
result that the entropy difference between the states is given by
Tl!.S = l!.U + Pl!.V- p Mi. (1 1-22)
From the statistical point of view, changes in the energy of an assembly, in its
volume, a nd in the number of particles result in changes in the total number of
possible microstates in which the system can exist. For example, iflhe energy U
o f the system in Fig. 11-4 is increased from 6< to 7<, the number of possible
micro,<;tates increases from 1532 to 2340 and the average occupation numbe!s of
each level change. (See Problem 11 -9.)
However, entropy is a n extensive property and the total entropy S of two
independent systems is the sum of(he entropies S 1 and S, of the individual systems:
S=S1 +S2 •
On the other hand, if!l1 and n,
a re the thermodynamic probabilities o f the systems,
a nd since for every microstate of either system the other may be in any one of its
possible microstates, the number n of possible microstates of the two systems is
the product of n, and n,:
n= n,n•. (11-23)
It follows that the entropy cannot be simply proportional to the thermodynamic
probability; and to find the form of the functional relationship between Sand n
such that the conditions above a re satisfied, we assume that Sis some unknown
function of !l, sayS = J(!l). The n since S ~ S 1 + S,, and !l = !l1!l0 ,

J (!l,) + J (!l.) = J(!l,!l., ).

Now take the partial derivatives of both sides of this equation, first with respect
to !l, withn, constant, and then with respect to !l, with !l, constant. Since J(!l1)
is a function of n, only, its partial derivative with respect to n, is equal to its
total derivative:

The partial derivative of J(!l2) with respect to !l1 is zero, since !l, is constant.
 324 STATtSnCAL TH ERMODYNAMICS 11-8

On the right side, the partial deriva tive of J (!l,O,), with respect to !l, equals
the total derivati ve of J(U,O,) with respect to its a rgument (0,0,), multiplied
by the patltial derivative of its argament with respect to !l,, which is simply the
consta nt 0 1• Then if we represent by J '(010,) the derivative of J(O,!l,) with
resi>cct to its argument, we have

dJ(Q,)- !l.,J'(!l,!l,).
dO,
In the same way,
dJ(!l,) •
- - - O,J (0 10 1).
dO,
It follows from these equations that
Q dJ(O,) = n dJ(Q,) .
1
dO, . .. dCl, '
a nd since 0 1 and 0. are independent , the equation can be satisfied only if each
side equals the same constant kn. Then for any arbitrary system,

Q dJ(Q) = k
dQ B•

dJ(O> = ks 0dQ ,
J(O) - k 8 In !l;
and hence
S- k 0 In !l. (11-24)
Thus the only fu nction of 0 which satisfies the condition that entropies are add/tire
while thermodynamic probabilities are multlplicatiue is the logari thm.
This equa tion provides the connecting link between statistical and classical
thermodynamics. T he nu merical value of the proportionality constant k n must
be chosen so that the classical and statistical values of the entropy will agree. We
shall show. in Section 11 - IS that kH turns out to be none other than the Boltzmann
constant k = R{N"'
From a statistical point of view the entropy of a system consisting of a very
large number of particles is proportional to the natural logarithm of the total
number of microstates available to the system. If we could prepare an assembly
so that energetically only one microstate is available to it, Q • I, In 0 = 0, and
the entropy would be zero. This system is perfectly ordered since the state of each
particle can be uniquely specified . If more energy states become available to the
system, Q becomes greater than I and the entropy is larger than zero. In this case
it is not possible to specify uniquely the state of each particle since the state of a
particle may be different when the system is in different microstates. Thus the
system becomes more disordered as more microstates become available to it.
 11-l! THE STATISTICAL INTER PRETATION OF ENTROPY 321

The entropy of the system may be thought of as a measure of the disorder of the
system.
This statistical interpretation of entropy allows additional insight into the
meaning of the absolute zero of temperature. According to the Planck statement
of the third law (Section 7-7) the e ntropy of a system in internal equilibrium
approaches zero as the temperature approaches zero. Therefore systems in internal
equilibrium must be perfectly ordered a t absolute zero.
Does the quantity k 11 In 0 have the other properties of entropy? We give some
qualitalitx' answers.
1. If there is a reversible flow of heat d' Q, into a system at a tempe ra tu re T, the
entropy of the system increases by dS ~ d' Q,/T. If the system is at constant volume
so that the wo rk in the process is zero, the increase d/) in internal energy of the
system equals d' Q,. But for an assembly of non interacting particles, the values of
the energy levels depend upon the volume; and if the volume is constan t, these
values do not change. If the energy of an assembly increases, more of the higher
energy levels become available to the particles of the assembly, with a corre-
sponding increase in the number of available microstates or the thermodynamic
probability 0 . Hence both Sand In 0 increase when the energy of the system is
increased.
2. The entropy of a n ideal gas increases in an irrttx'rsiblt free expansion fro m a
volume V1 to a volume V,. There is no change in internal energy in the process,
a nd no work is done, but the permitted energy levels become more closely spaced
because of the increase in volume. For a constant total energy, more microstates
become available as the spacing of the energy levels decreases, and again both S
and In 0 increase in the irreversible free expansion.
3. In a reversible adiabatic expansion of an ideal gas, the entropy S remains
constant. There is no hea t flow into the gas, and the work in the expansion is done
at the expense of the internal energy, which decreases in the process. If the spacing
of the energy levels did not change, a reduction in internal energy would result in
a smaller number of available microstates with a corresponding decrease In In 0,
but because of the increase in volume the energy levels become more closely
spaced, and the resulting increase in In 0 just compensates for the decrease arising
from a decrease in internal energy. The result is that In O, like S, remains constant.
Many other examples could be cited, and it turns out in fact that complete
agreement between thermodynamics a nd statistics results from the assumption
that the entropy S , whose change dS is defined in the rmodynamics by the relation

dS- d'Q,
T •

has its statistical counterpart in the logarithm of the the rmodyna mic probability
n of a n ass~mbly of a very large number of pa rticles, or in the logarithm of the
321 STATISTICAL THERMOOYNAMICS 11 - 8

tota l number of microstates available 10 the assembly. Thus if S = k 0 Inn, the

entropy difference between two neighboring Slates of an assembly is dS -
k 11 d(lnfi).
Add itional insight into the connection between statistical and classical therm o.
dynamics can be gained by considering two neighboring states or a closed system, in
which the values of the internal energy U, the energy levels ,,, and the average
occuft'tion numbers iV1 are slightly different. Since the energy U is given by I,c1N"
the energy difference between the states is then

01-m
that is, the difference in energy results in part from the differences dF/1 in the average
occupation numbers, and in part from the differences d<1 in the energy levels.
If the values or the energy levels are function.s of son1e extensive parameter X,
such as the volume V, then
(1 1- 26)
and

~N1 d•1 - [:~N~ ~]dx.

Ltt us define a quantity Y as
d.,
y'" -~N1 dX ' (11-27)
Then
IN1 d,,- -YdX.
I
(11- 28)

If, for example, the parameter X is t~e volume V, the quant ity Y is the pressure P and

YdX- PdV.
The energy difference dU is then

dU -I,, dN, - y dX.

'
For two states in which the va lue or the parametor X is the same, dX = 0, and

dUx =I •1 dN1•
'
:'""principles or thermodynamics lead to the result that whon X is constan t,

dUx- TdS,
and hence
(11- 29)
Thus the equation
 11-9 THE BOSE -EINSTEIN DISTRIBUTION FUNCTION 327

I
is the statistical form of the combined first and second law of thermodynamics for a
closed system:
dU- T dS- YdX.
If the system is taken from one state to the other by a rtcrrsiblt process, then
T dS- d'Q,, and Y dX - d'W,.
Hence in such a process,
dU - d'Q, - d'W,
and
f 'I dl'll - d'Q,. f Nl d.l - - d' W,. ( 11- 30)

It is sometimes assumed that the sum

d'Q into the system and sum :r, :r,
<1 dl'l1 Is always equal to the heat flow
N1 d<1 is always equal to the negative of the work
-d'W. We see that this is the case for a "'"'"iblt process only, and only for such a
process can we identify the sums in Eq. (II-2S) with the heat flow and the work.

11- 9 THE BOSE- EI NSTEIN OISTRIBUTION FUNCTION

If a system consists of only a relatively small number of particles, as in Fig. 11-4,
the average values of the occupation numbers of the energy levels can be calculated
without much difficulty, when the total number of particles and the' total energy
are fixed . When the number is very large, as in the statistical model of a macro-
scopic system, direct calculations are impossible. We now show how to derive
a general expression for the average occupation numbers when the total number of
particles is very large. Such an expression is called a distribution function. The
procedure is firs t to derive a general relation for the relative values of In fi for
two systems having the same set of energy levels, but in the second system the
number of particles is less than that in the fi rst by some small number n, where
n « N, and in which the energy is less than in the firs t by n•,. where •, is the energy
of some arbitrary level r. Thus if unprimed symbols refer to the first system and
primed symbols to the second system,
N ' = N- n, U' == U- ,,,. ( 11-31)
These conditions can always be met, since we can control independently the
number of particles in the system, and its energy. The difference in the values of
kn In fi is then equated to the entropy difference between the systems, using Eq.
( 11-24).
The only way in which Eqs. (1 1-3 1) can be satisfied is that in every macro-
state of the primed system the occupation numbers of all levels, with the exception
of level r , are the same in both systems, while the occupation number of level r
in the primed system is less than that in the unprimed system by the number n.
That is, to satisfy Eqs. ( 11- 31) we mu~t have in every macrostate k,

Nj ;.. NdJ ,& r), N; • N,- n. ( 11-32)

 328 STATISTICAL THEAMOOYNAMICS 11-9

·-
f/ t • 61 •
I JO .,...!!-
0.()41
N, N •6
u •6•
s[ 0.018 n •lll2
41 O.lOl fJ•l

lj 0.410

lj
...
0.8.10

li
o["
. '..' ..
.. .. ·. ...
1.60

ll)

··- 63 Ill Ill ISO 90 210

(1)
ISO 100 216 13r'28'

-
4
I 2 ) 4 l 6 7 8 L ,.!L ;;;,
0. 129
N' •l
( f . 4t

J . 0.2S8 Oj • 341

2 .. . 0.6ll ,, •l

.
0
I

., ..u-
.. .
4l
... . . . .
90 60
t'
108
• 'f----

4$
'--
I
ll6
.40

(b)

Fig. I 1-9 (a) The pos.>ible macrostates of an assembly of 6 particles obeying

B·E statistics when U • 6<. (b) The poS.!ible macrostates when one particle
is removed from level 2 o f the assembly of part (a). The thermodynamic
probability of each macrostate is given at the bottom and the average occupa·
tion number of each level is printed on the right of the diagram.

The result is equivalent to the removal of n particles from level r in the unprimed
system, without changing the occupation numbers of the other levels.
We censider first a system obeying the Bose-Einstein statistics, and illustrate
the relation between corresponding macrostates by taking as an example of the
unprimed system that of Fig. 11-4, shown again in part (a) of Fig. 11-9. The
number of particles is N - 6, the energy U = 6<, and we let n have its smallest
possible value, n = I. The number of particles in the p rimed system is N ' =
N - I ~ S, and level 2 has been selected as t.he a rbitrary level r so that the energy
of the p rimed system is U' = U - 2• = 4 <.
 11-9 THE BOSE·EIN STEIN DISTRIBUTION FUNCTION 32,

The only possible macrostates of the primed system are shown in part ·(b) of
Fig. 11-9. There can evidenrly be no macrostates of the primed system corre·
sponding to a macrostate of the unprimed system in which level 2 is unoccupied.
Thus there are only five possible macrostates, and it will be seen that in each of
these the occupation number of level 2 is one less than in the corresponding
macrostate of the unprimed system, the occupation numbers of all other levels
being the same in borh sysrems.
The thermodynamic probability 71'"• of macrostate k in the unprimed system is

(11- 33)

In the primed system,

1r' _ fi (g, + Nj; - I )!
( 11-34)
•• - , (g, - I)! Nj•l .

The double subscript rk means that tr;. is t~e thermodynamic probability of

macrostale kin the primed syslem, and Ihat level r has been selected as the arbitrary
level from which o ne particle has been remov~d. The double subscript jk means
that N,. and Nit are, respectively, the occupation numbers of level ) in macrostate
k, in the unprimed and primed systems.
The fact that rhere are no macrostates in ihe primed system corresponding to
states in the unprimed sysrem in which level r is unoccupied is equivalent to stating
that for such macrosrates the thermodynamic probability if'";~ is zero. But if
N,. - 0, then N;. - 0 - I - -I, and the rth term in the product in Eq. ( 11-34)
becomes
(g, - 2)!
(g,- 1)!(- 1)! (g,- I)( -1)1

Hence in order that ir;. shall be zero, and provided that g, > I, we must adopt
the convention that (- I)! ~ oo. For a more general discussion, see Appendix C.
The ratio of thermodynamic probabilities is

In all levels except level r, Nit = N,.

so thar allrerms in the product above will
cancel bel ween numeraror and denominator, with the exce"prion of level r in which
N;t = N,-. - I. Therefore, since

and
(g, + N,. - I)! - (g, + N;.J! = (g, + N;.)(g, + N;. - 1)!,
 330 STATISTICAL THERMOOYNAMICS 11-9

or
N,•ir• = (g, + N;.)-tr;.;
and summing over all macrostates,

"i,N,.~:irt ~: g,.I,tr;. + 2, N;~:ir;1t.

• • •
The term on the left, from Eq. (I J-8), equals Ji/,0. On the right, the term
g,!o-tr;. equals g,O; a nd the last term equals R;o;. Therefore
R,O - (g, + R;Jo;
and
Jil, o;
g, + R;- a· (11-35)

In a macroscopic system in which the occupation numbers are very large, the
removal of o ne part icle from a level will make only a rela tively s mall c hange in the
average occupation number of the level, and to a good approximation we can set
R; - Jil, so that -
Jil, o;
( 11-36)
g, + R, = fi.'
Taking the natural logarithms of both sides, we have

In -
N
- '- = ln-.
o;
g,+R, o
But

In ~ ..; In o; - In 0;

and since by Eq. (11- 24), S- k 0 In 0,

In~- = S' - S = llS (11-37)

g, + R, kn kn ·
From the principles of thermodynamics, the entropy differe nce llS between
two states of a no nisolated open system in which the volume (or the appropriate
extensive variable) is constant is related to the energy diffe rence llU, the difference
AN in the number of particles, a nd the temperat ure T, by Eq. (8-1J) :
TllS =- llU- pllN,
 11-10 THE FERMI-DIRAC DISTRIBUTION FUNCTION 331

where p. is now the chemical potential per particle. For the two states we are con-
sidering,
AU= -•., AN= -1,
and hence

Then from Eq. {ll-37), since level r was arbitrarily chosen and could be any ·
level j,
In ___!!_t_ = P. - ., ;
g1 + N, k0 T
and
8' + N, = .!t. + 1 = exp •, - P.
N, N, k0 T '
which can be written as

N, = ------~----- {11-38)
exp ( '' - P.) - I
k8 T

This equation i.s.t.he Bose-Einsttin distribution function. It expresses the average

occupation number per state in any level j, N1/g1, in terms of the energy ., of the
state, the chemicat potential p, the universal constant kn, and the temperature
T. Of course, to apply the equation to a particular system we must know the ex-
pression for the energies <1 of the permitted energy levels, and for the chemical
potential p. Another derivation of Eq. {11-38) is found in Appendix D.

11- 10 THE FERM t·DIRAC DISTRIBUTION FUNCTION

To derive the distribution function in the F-D statistics, we agai n consider two
assemblies in which the nu mbers of particles are respectively Nand N' = N - I.
In any pair of corresponding macrostates, Ni• = N,. in all levels except an arbitrary
level r; and in level r, N;. = N,. - I. The corresponding energies are U and
U' = U- <,.
Part (a) of Fig. 11- 10 is the same as Fig. 11-6 and shows the possible macro·
states of an assem bly of N = 6 particles and U = 6<, for an assembly obeying
the F-D statistics and in which the energy levels are equally spaced and g 1 = 3
in each level. Part (b) is the corresponding diagram for an assembly of N' = 5
particles and one in which level 2 has been chosen as the arbitrary level r so that
U' = U - 2• = 4<. Again it will be seen that in every pair of corresponding
macrostates the occupation numbers are the same in all levels except level 2, and
that in this level N;. = N,. - I.
 332 STATISTICAL THERM ODYNAM ICS 11- 10

·- .
t.J t •4
I N,
O. IZJ
N •6

u. 6•
l •
t/ •· •
I Nj N•S
U'. 4t

ll'. 39
0.4 94 n -n 0.231

... • u s ,,.J 0.846 IJ • J

.. • 1.62

... ... .. .·.

I • •1.1.1
..
0

···- 9 21 9
(o)
• lSI
27 ...,. . (b)
• l ll
27

Fig. 11-10 (a) The possible macroslatcs of an assembly of 6 particles obeying F·D
slalistics when U = 6<. (b) The possible macrostatcs when one particle is removed from
level 2 of the assembly of part (a). The lhcrmodynamic probability of each macrostate is
given at the bottom and the average occupation number of each level is prinled on I he
right of the diagram.

The thermodynamic probabilities of corresponding macrostates in the un·

primed and primed a ssemblies are
il/' = rr g,,
• 1 (g 1 - N,.)! N"!'

if";. = rr g,, .
1 (g 1 - Nj.)l Nj.l
Then
if";.= I1 (g l - N,.)l N,.!
if",. 1 (g 1 - Nj.)! Nj, !'
which aftef cancellation reduces to

Yr;. N,t
if", = g, - N;,
or
N,,if". - (g, - N;,)tr;..
Summing o ver all values of k, we have

I N,,if", = g, I -rr ;, - I N;,r ;,

• • •
and

(JJ -39)
 11- 12 COMPARISON OF DISTRIBUTION FUNCTIONS 333

Here we can let FJ; = FJ, since if the states a re degenera te enough, N, and N; can
be much la rger than one. By the same reasoning as in the B-E statistics
!!..t= ____:___ (11-40)
g, exp ( •' - "') + I •
k8 T
which is the Ftrmi-Dirac distribution f unction. It differs from the B-E distribution
in that we have + I in the denominator instead of -I.

11-11 THE CLASSICAL DISTRIBUTION FUNCTION

I n many systems of indisti nguishable particles, the average nu mber ofparticles

!iJ1 in a level is very much less than the number of states g1 in the level, so that the
average number of particles per state, liJ1fg1, is very small. The denominator i n
Eqs. ( 11-38) and (I 1-40) must then be very large; we ca n neglect the I ; and both
the B-E and F-D distribution functions reduce to
FJ, iJ - . ,
-:a: exp - - , (11-41)
g, k0 T
which is the classical distribution function.

11-12 COMPARISON OF DISTRIBUTION FUN CTIONS FOR

INDISTINGUISHABLE PARTICLES
The distribution functions for indistinguishable particles can all be represented by
the single equation,
!l.!.= ----=---- (11-42)
g, exp (•' - ~-') +a•
k0 T
where a = - I in the B·E statistics, a = +I in the F·D sta tistics, and a = 0 in
the classical statistics.
The curves in 'Fig. II-II are graphs of the average number of particles pe r state,
N1/g1 , at a given temperature, for the B-E and F-D statistics, plotted as functions
of the dimensionless quantity (<1 - p,)/k 11 T. (The e nergy therefore increases toward
the ri ght.) The ordinates of the curves have a meaning, of course, o nly at those
abscissas at which the energy <1 has some one of its permitted values. When
N1/g1 is very small, the B·E and F-D distributions very nearly coincide, and both
reduce to the classical distribution.
Note that when •1 = p,, the value of !iJ1/g1 in the B-E statistics becomes in-
finite, a nd for levels in which • 1 is less than p, it is negative and hence meaningless.
Tha t is, in this statistics, the chemical potential must be less than the energy of
the lowest permitted energy level. The particles like to concentrate in levels for
which •1 is only slightly greater than p,.
 334 STATISTICAL THERMODYNAMICS 11-13

-) -2

Fig.Jl-11 Graphs of the Bose-Einstein, Fermi-

Dirac, and classical distribution functions.

In the F-D statistics, on the other hand, all levels a re populated down to the
lowest and as <1 decreases, R,/g, approaches I. That is, the low-energy levels are
very nearly uniformly populated with one particle per state.
The curve for classical statistics has no meaning except when (•, - Jl)/k T is
large. It is drawn on Fig. I I-I I for comparison only. If the ordinate of Fig. 11- 11
is taken as R1/Ng1 instead of R,/g,, this curve is the distribution function for
M-B statistics which is developed in the next section.

11-13 THE MAXWELL-BOLTZMANN DISTRIBUTION FUNCTION

The distribution function in the M-B statistics is de rived in the same way as in the
B-E and F-D statistics. Part (a) of Fig. 11 - 12 is the same as Fig. 11- 8, in which
the dots represent the occupation num bers of an assembly of N = 6 particles and
of energy U = 6<. Pa rt (b) shows the possible macrostates of an assembly of
N ' = N - I = 5 particles, and in this assembly level 2 has been chosen for the
arbitrary level r so that U' = U - 2< = 4<. . The only possible macrostates of
the primed assembly a re those in which level 2 is occupied in the unprimed
assembly. In any pair of corresponding macrostates, the occupation numbers are
the same in all levels except level 2; and in level 2, N;. = N.. - I.
The thermodynamic probabilities of corresponding macrostates in the un-
primed and primed assemblies a re
gi'll
tr. = N! IT _!_.
, N1 l
N'
ir~ = N'! fi ~.
'Nj!
 11- 13 THE MAXWELL-BOLTZMANN DISTR IBUTION FUNCTION 335

k• I 2 10 II N, N •6
•J~• -6
r--r-
Jo.ou U •IH

0 .065 Cl •IJ86 JC 3~

0 .195 fJ•l

I
01-;-;·• • - • • ··;

• J .l ) - • 18 • • 90 • • 90 • I~
1•1

(b)

Fig. 11- 11 (a) The possible macrostates or an assembly or 6 particles obeying M -B

statistics when U - 6<. (b) The possible macros tales when one particle is removed
from level 2 or the asstmbly of part (a). T he tht rmodynamic probability of each
macrostatt is given atth< bouom and the avtrage occupation number or t ach level
is printed on the right or the diagram.

T hen

which simplifies to

or
 338 STATISTICAL THERMODYNAMICS 11- 14

Summing over all macroslales, we have

tv,. n; (11-43)
Ng, =a·
and by lhe same procedure as before,

IV,/N 1 p - <
- -crexp--, (11-44)
g1 knT
which is Ihe Maxll'tll· Boltzmann distribution Junction. II differs from I he classical
d islribulion funclion, which is somelimes referred 10 as I he "correcled" Bollzmann
funca ion , in that lhe numeralor on the left is the average fractional number of
particles in level}, IVJN, so. thalthe left side is the fractional number of parlicles
per state in any level.

11-14 THE PARTITION FUNCTION

The distribution function in the Maxwell-Boltzmann statisaics can be wrillen

IV, L)g,e xp k- •,T .

= N(exp kuT
11
_Since ! 1 tv, .. N, and the chemical potential p does neil depend on j, it follows
tha t

! IV1 =- N
.. N{expL)! g 1 exp - • , .
1 kuT 1 kuT
The sum in the last lerm is called the partitlon fimr!ion or sum outr stat~s a nd
will be represenled by Z . (German Z11standssumm~) Oaher leite rs are oflen used.

( 11-45)

The partilion funclion depends only on the lemperalure T and on I he pa ra m-

elers I hat determine 1he energy levels. II follows from the 1wo preceding equations
that in lhc M·B slalislics,
IJ I
exp kn T = Z' (11-46)

and hence the M-B dislribulion funclion can be wrillen

IV, N - •
- =-exp -1 . ( 11-47)
g1 · Z knT
Thus in a given system, the average number of pa rticles per slate in any level
decreases exponenlially wilh the energy <1 .of lhe level : and lhe lower the !em-
perature T, the more rapid is Ihe rate of decrease.
 11-15 THERMODYNAMIC PROP ERTIES OF A SYSTE~ 337

The cla.ssical distribution function can be written

tv,= (exp L ) g,exp -•,,

kaT kaT
and summing over aU values of j, we have

I tv, • JJ )
N = ( exp- I g1 exp- -
•, .
1 kaT 1 koT
Then if the partition function Z is defined in the same way as in the M-B
statistics, we have
JJ N
exp ksT = z' (11--48)

and the classical distribution function can be wriuen

-tv,
g
= N-Z exp-,
- •
k T
1
(11--49)
1 0
which has the same form as the M-B distribution.
Because of the form of the B-E and F-D distribution functions, these cannot
be expressed in terms of a single-particle partition function, and we shall discuss
them later.

11- 15 THERMODYNAMIC PROPERTIES OF A · SYSTEM

Tlfe importance of lhe partition function Z is that in Maxwell-Boltzmann and
classical statistics, all the thermodynamic properties of a system can be expressed
in terms of In Z and its partial derivatives. Thus the first step· in applying the
methods o f statistics to such a system is to evaluate the partition function of the
system.
It will be recalled tha t all thermodynamic properties of a system are also com-
pletely determined by its characteristic ''Illation; tha t is, the Helmholtz funct ion
expressed in terms of X and Tor the G ibbs function expressed in terms of Y and
T. Here X and Y stand for some related pair of variables such as the volume V
and the pressure P.
Thus we begin by deriving expressions for t)le Helmholtz and Gibbs functions
in terms of In Z. As shown in Section 8--1, these functions are related to the
chemical potential p by the equation

p.=- oG)
(oN 1', Y
c: (ooNF)
-
T . x'
(II-SO)

For a system obeying M-B statistics, the chemical potential of the system ·is related
to the partition function by Eq. (1 1--46):
f' = -k 0 T in Z. {I l-SI)
 3:18 STATISTICAL THERMODYNAMICS 11-15

In classical statistics, the chemical potential is given by Eq. (11-48):

p .. - k0 T(Jn Z- InN). (11-52)
The partition function , Z = Eg1 exp ( -•1/k,.T), is a function of the temperature
of the system and of the paramete~ that determine the energy levels of the system
(such as the volume Vor the magneticintensity Jl"). T hus Eqs. (I I- 51) and (II-52)
express p in terms of X or Y.
Consider first a system of indistinguishable particles obeying the classical
statistics and one in which the energy levels are functions of an extensive parameter
X. Then the partition function is a function of X and T, and as these are the
"natural" variables of the Helmholtz function F, we have from Eqs. (II-50) and
(II-52),
CIF\
( CIN/T.x = -k0 T(Jn Z -In N). (II-53)

The right side of this equation is constant when X and Tare constant. Integrating
at constant X and T yields
Fa - Nk0 T(ln Z- InN+ 1), (11-54)
since JN InN dN- N InN- N. Equation (II-53) would be satisfied if any
functionf(T, X) were added to the right side of Eq. (II-54), but since F must be
zero when N = 0, it follows thatf(T, X) = 0. Equation ( 11-54) is on expression
for Fin terms of N, T, and X; therefore all the thermodynamic properties of the
system can be determined by tf1e methods of Section 7- 2.
The entropy Sis given by S = - (CIFfCIT)K.x so that

S = Nk 0 T ( - z)
CI In- + Nk,.(lnZ -InN+ 1). (II-55)
CIT x
Since U =F + TS, the internal energy is

u= Nkur' ( CIInZ) . (II-56)

CIT x
The expression for the entropy can now be rewritten as

S - !!. + Nk 0 (1n Z - In N + 1). (11 - 57)

T
The intensive variable Y associated with the extensive variable X is given by
Y = - (oFfoX)N.'r • so that
Y = Nk
"
r( ax IT'
11 In Z\ (11- 58)

which is the equation of state of the system, expressing Y as a function of N, T,

and X. Thus all thermodynamic properties of this system can be determined if Z
is known as a function of X and T.
 11-15 THERMODYNAMIC PROPERTIES OF A SYSTEM 339

For a one-component system, the Gibbs function G = pN, so that from Eq.
(II-52)
G = -Nk 0 T(InZ -In N). (11-S9)

But in general for the variables X and Y,

G = U- TS- YX = F + YX,
and
G-F - YX.
From Eqs. ( II -S4) and (II-S9),
G-F=Nk 11 T,

so that for any system obeying the classical statistics and in which the energy levels
are functions of a single extensive parameter X,
rx ... NknT. (11-60)
In the special case in which the pnrnmeter X is the volume V and Y is the
pressure P,
PV= NknT.
This is the equation of state of an ideal gas as derived from kinetic theory, pro-
vided that the universal constant k 11 , which was introduced earlier only as the
proportionality constant in the equation S = k 11 Jn U. is equal to the Boltzmnnn
constant k - R/N.,. Since k 11 is a tmitwsal constant, which in this special case is
equal to R/N,,, it must equal R/N ., regardless of the nnture of an assembly. In
the future we shall, for simplicity, drop the subscript Band write simply S = k In 0.
It is at first surprising that we obtain only the ideal gas equation of state.
However, the partition fu nction can only be given by the sum over single particle
states when the pnrticles do not interact. This is the same condition needed to
derive the ideal gas law from kinetic theory.
In t~rms of this notation, the expressions for the thermodynamic properties
of a system obeying classical statistics and a system in which the energy levels are
determined by the extensive parameter X are given by I
F - -NkT( In Z - InN + I), (11-61)

u - Nkr'(a In
ar -"
z) ' (11-62)

S= Q + Nk(ln Z- In N + 1), ( I 1-63)

T
and
y = NkT(o ln
oX
Z).
T
(11-64)
 340 STATISTICAL THERMODYNAMICS 11- 15

It is left as an exercise (Problem 11-30) to show that for a system of dis-

tinguishable particles obeying M-B statistics and in which the energy levels a re
determined by an extensive parameter X, the expressions for U and Y arc un-
changed, but the expressions for F and S are
F--NkTlnZ (11-65)
and
S =!!. + Nk ln Z . (11-66)
T
These expressions differ from those for indistinguishable particles by a term pro-
portional toN ln N - N. (See Problem 11-31 ).
As a second example, consider a system of distinguishable particles obeying
the M-B statistics and for which the energy levels are functions of an intensive
parameter Y. Then Z is a fu nction of Y and T ; and since these arc the "natural"
variables of the Gibbs funct ion, we have, from Eqs. (II- 50) a nd (11 -SI),

(Efl)
oN T .Y
~ - kT In Z. (11- 67)
I
The right side of this equation is constant when Tand Y arc constant. Integrating
at constant T and Y yields
G = -NkTln Z . (11-68)
The arbitrary function g(T, Y) which should be added to the right side of Eq.
( 11-68) is again zero since G ~ 0 when N ~ 0. This equation appears at first to
contradict Eq. (1 1-65) since F ,e G. However, Eq . ( 11 -65) is derived for a system
in which the energy levels arc functions of an extensive parameter X, whereas
Eq. (11-68) applies to a system in which the energy levels depend upon an intensive
parameter Y.
The entropy is now given by S- - (oGfo T)_,·,y, and hence

s= NkT( 0 In z) + Nk ln z. (11 -69)

ar y

The enthalpy H equals G + TS, so

H = Nk T' ( il In
ar z)y ' (11-70)
and Eq. (11-69) can be writtcr

S= TH + Nk lnZ. (11-71)
The equation of state is given by

X= (ioTlG).v.7 = - NkT( olnZ).

ilY T
( 11 -72)
11-15 THERMODYNAMIC PROPERTIES OF A SYSTEM 341

If the parameter Y is the intensity of a conservative field of force, the only

energy of the particle is its potential energy (gravitational, magnetic, or electric).
The internal energy of the system is then zero, and its total energy E is its potential
energy £ 1, only. If X represents the extensive variable associated with the mtensiv~
variable Y, the potenlial energy E, - YX. Then since the enthalpy His defined
asH= U + YX, and U = 0, it follows t hat
E= E,- H ,
and Eqs. (11- 70) and (11 -71) can be written

F- NkT' (i1ln
i1T
z) y'
( 11-73)
and
E
S = T + NklnZ. (11-74)

It has been assumed thus far in this section that the energy levels were functions
either of a single extensive variable X or a s ingle intensive variable Y. W e now
consider the more general case of a multivariab/e system in which the energy levels
are functions of more than one independent variable. We restrict the discussion to
systems whose energy levels are functions of two variables only, one of which is
an extensive variable X1 while the other is an intensive variable Y1 , which we
consider to be the intensity of a conservative field of force.
If the system is described by either the Maxwell-Boltzmann or classical
statistics, we can still define the partition function as

Z = 1, gtexp (-•t).
kT
The only difference is that the •js are now functions of both X, and Y1 , and the
partition function is a function of T, X,, and Y1 . Since the system has both an
internal energy U and a potential energy £ 11 - Y,X,, its total energy E is

E- u+ E,- U + Y1 X,,
and we therefore make use of the generalized Helmholtz function F*, defined by
Eq. (7-34) as
F• 11 E- TS- U- TS + Y,X 1 •
The chemical potential is now
f'- (ilF:'\ .
ilN}T.X,.Y,
If the system obeys the classical statistics,

I' - -kT(ln Z - In N),

 342 STATISTICAL THERMODYNAMICS 11- 15

and, integrating at constant T, X 1, Y,,

F* "' -NkT(In Z -InN + 1), (11-75)
setting the arbitrary function of X1 , Y,, and Tequnl to zero as before.
The variables Y1 and X, , associated with the variables X 1 and Y,, are given by

Y, = -(oF*)
oX, .V.T.Y1
= NkT(oln
oX,
z) .
T.Y,
(11-76)

X,=(oF•) =-Nkt(o lnZ) . ( 11-77)

oY, .v.T.X, oY, T.x ,
The system thus has 111'0 equations of state, expressing Y1 and X, in terms of N,
T, X,. and Y,.
The entropy S is

s = -(oF*) = Nkr( 8 In 2) + Nk(ln z - InN + 1). ( I 1-78)

aT .v.x,.Y, aT .v,.Y,
The total energy E equals F* + TS, so

E = NkT' ( o In z) •
oT x,.Y,
(1 1-79)
and hence
S = E
T
+ Nk(ln Z - InN + 1). ( 11 -80)

If the system obeys the Maxwell-Boltzmann statistics,

I'= -kTln Z;
and by similar reasoning,
F• = -NkTln Z (11-81)
The variables Y1 and X, are again given by Eqs. ( I 1-75) and (I 1-76). The entropy
is
0 2
S = NkT( In ) + Nk In Z. (11-82)
aT x,.Y,
The total energy is
E = Nk r ' ( a In
oT x,.Y,
z) . (11-83)
so one can also write
S= E + NklnZ. ( 11- 84)
T
In either statistics, the potential energy £ 1, = Y,X, and the internal energy U is
U=E-E,,=E- Y,X,. (11-85)
Specific examples of the general relations derived in this section will be dis·
cussed in the next two chapters.
 PR OBLEMS 343

PROBLEMS

11-1 Using quantum mechanics, show that the energy levels or a one-dimensional
infinite square well or width L arc also given by Eq. (1 1-3).
11- 2 (a) Tabulate the values or the quantum numbers n., n,, n, for the twelve lowest
energy levels or a free part ide in a container or volume V. (b) What is the degeneracy!
or each level ? (c) Find the energy or each level in units or h2/8m Vli>. (d) Are the energy
levels equally spaced?
Jl-3 Calculate the value or n 1 in which an oxygen atom confined to a cubical box I em
on a side will have the same energy as the lowest energy available to a helium atom con-
tined to a cubical box 2 x 1()-••
m on a side.
Jl-4 Five indistinguishable particles are to be distributed among the four equally spaced
energy levels shown in Fig. 11-2 with no restriction on the number or particles in each
energy state. If the total energy is to be 12<" (a) specify the occupation number or each
level for each macrostate, and (b) find the number or microstates for each macrostate,
given the ene'l)' states represented in Fig. I 1- 2.
Jl-5 (a) Find the number of macrostates for an assembly of four particles distributed
among two energy levels one or which is two-fold degenerate. (b) Find the thermody·
namic probability or each macrostate if there is no restriction on the number or particles
in each energy state and the particles are indist inguishable, (c) distinguishable. (d) Cal-
culate the thermodynamic probability or the assembly for parts (b) a nd {c).
11-6 In the poker game stwn-rord stud, seven cards are dealt to each player. He makes
the best hand out or five of those cards. The cards are well shuffled between each deal.
{a) How many different seven-card hands can be made in a deck of 52 cards? (b) If there
are fou r players, how many different ways can the cards be dealt if the players are dis-
tinguishable? (c) How many different five-card hands can be ntdde from a seven-<:ard
hand?
11-7 For the example illustrated in Fig. 11-4, find (a) the thermodynamic probability
lrt of each macrostate, {b) the total number of microstates of the assembly 0, (c) the
average occupation number of each level, and (d) the su m of the average occupation
numbers.
Jl-8 Do Problem 11 - 7 for a system or seven indistinguishable particles obeying B-E
statistics and having a total energy U - 6<.
11-9 {a) Construct a diagram similar to Fig. 11-6, but having eight energy levels. Show
the possible macrostatcs of the system if the energy U - 7< for six indistinguishable
particles, obeying B-E statistics. {b) Calculate the thermodynamic probability of each
macrostate, and (c) show that the to tal number of possible microstates 0 is 2340. (d)
Find the average occupation number of each level.
11-10 (a) Suppose that in the F-D statistics, level j includes three sta les (I), (2),1(3), and
two particles a and b. If the particular sequence of numbers (1), (2), and (3), is selected,
write down the possible different sequences or leuers and numbers, and show that th is
agrees with Eq. (11- 15). {b) How many different sequences of numbers are possible?
(c) What is the total number of different possible sequences of Jeuers and numbers?
 344 STATISTICAL THERM ODYNAMICS

11-11 Show tha t in the Fermi-Dirac statistics, irlevel j is fully occupied with one particle
per state, 11·, - I and there is only one way or distributing the particles among the energy
states or that level.
11- 12 Do Problem 11- 9 for six indistinguishable particles obeying F-D stat istics. In
this case 0 - 162.
11- 13 Do Problem 11-9 for six distinguishable particles obeying M-B statistics. In
this case 0 - 5.17 x 10'.
11-14 There are 30 d istinguishable particles distributed among three nondegenerate
energy levels la beled I, 2, 3, such that N 1 - N, - N,- 10. The energies or the levels
are E"1 - 2 eV. c-1 .. 4 eV, C'3 - 6 eV. (a) Ir the change in the occupation number of
level 2, •W, - -2, find •IN1 and •IN, such that •IE - 0. (b) Find the thermodynamic
probability of the macrostate before and a fter the change.
11- 15 Six Idistinguishable particles are distributed over three nondegenerate energy
leve ls. Level I is at zero energy; level 2 has a n energy <; and level 3 has an ene rgy 2<.
(a) Calculate the total nu mber of microstates for the system. (b) Calculate the number of
mjcrostates such th1U lhere are three particles in level 1. two in level 2, and one in level 3.
(c) Find the energy of the distribution for which 11 "• is largest. (d) Calculate the total
number of microstates if the total ene rgy of the six particles is 5<.
11- 16 Five particles are distributed among the states of the four equally spaced ene rgy
levels shown on Fig. 11-2 such that the total energy is 12<1 • Calculate the thermodynamic
probability of each macrostate and the average occupation number of each level if the
particles obey (a) B-E, (b) F-0, (c) M-B statistics.
11-11 Calculate the change in the entropy of each of the systems illustrated In Figs.
11- 4, 11 - 6, and 11 - 8 when a n add itional energy level is available to the pa rticles and the
total energy is increased to 7<. [See Problems 11- 9, 11-12, and 11-13.]
11- 18 The internal energy of the six ind istinguishable particles of Fig. 11- 4 is increased
reversibly from 6< to 7< without work being done, but only the levels up through level 6
can be occupied. (a) Show explicitly that d'Q, - _!1 ., d/il1 and (b) find the increase in
the entropy of the system.
11- 19 (a) Construct a diagram similar to part (b) of Fig. 11-9, bu t in which level 3 is
selected as the arbitrary leve l r so that U' - 6< - 3• - 3<. Note that every possible
macrostate or the primed system corresponds to a macrostate or the unprimed system
and that with the exception of level 3 the occupation numbers of all levels are the same
in each pair or corresponding macrostates. (b) Ho w many possible macrostates are there
for the primed system? (c) How many microstates? (d) Calculate the average occupation
number of the levels of the primed system. (e) Use Eq. (1 1-35) to calculate the average
oocupation number or level 3 or the primed system. (f) Calculate the change in the
entropy of the unprimed system upon removing one particle from level 3.
11- 20 Fill in the steps of the derivation of (a) Eq. (11-39) and (b) Eq. ( 11-40).
11-11 (a) Construct a diagram similar to part (b) of Fig. 11- 10 but in wh ich level 3 is
selected as the arbitrary level r so that U' - 3<. (b) Calculate the number of microsta tes
a va ilable to the primed system. (c) Calculate the average occupation number of the levels
or the primed system. (d) Use Eq. ( 11 -39) to calculate the average occupation number of
level 3 or the pri med system. (e) Calculate the change in the entropy or the unprimed
system upon removing one particle from level J.
 PROBLEMS 345

11- 22 Show fhaf Eq. ( 11- 13) for ,-,..~ and Eq. (1 1-17) for tr ,.., bofh roduce fo

( 11 -86)

in fhe limif fhaf g 1 » N1• This is fhe fhermodynamic probnbilify or a sysfem obeying
classical sfafiSiics.
11- 23 By a mefhod similar fo fhaf in Seclion JJ -9, show fhaf Eq. ( 11-86) of fhe pre-
vious problem leads fo fhc d isfribufio n funcfion of Eq. (1 1-41).
11- 24 Show fhaf Eq. (11-13) for ,..n·R• Eq. (11- 17) for.,.,._, and Eq. (11-86) (Pfoblem
11-22) for classical sfafisfics can all be r<presenfed by
..,. _ ITgj(g1 - a)(g1 - 2a) · · · [g1 - (N1 - I )a)
I Nil •
where a has fhe va lues given in Secfion 11- 12.
11-25 Fill in fhc sfeps of fhe derivafion of fhc Maxweii-Bollzmnnn disfribufion funcfion
done in Secfion 11- 13.
11-26 Derive fhc Maxweii-Bollzmann disaribufion funcfion by fhe mefhod of Secfion
11- 13 buf assume fhafll pnrlicl., are removed from fhc level r of fhe unprimed system,
where n « N.
11- 27 (a) Consfructa diagram similar to part (b) or Fig. 11 - 12 but one in which level 3
is selecfed as fhe arbitrary level r so that U' - 3<. (b) Calculafe the number of micro-
Sfafes available fO fhe primed sysfem. (c) Calculafe fhe average occupafion number of fhe
le vels of the primed system. (d) Calculafe fhe change in fhe enfropy of the unprimed
system upon removing one particle rrom level 3.
11-28 Subsfifufe fhe Maxweii-Bollzmann disaribufion funcfion info Eq. (11-29), the
expr<SSion for fhe enaropy change or a sysfem in a reversible process, fO obfain

R,
S - -k ~ N1 1n-.
7' K•
11- 29 Seven disfinguishable panicles art dislribufed over fwo energy levels. The upper
level is nondegenerafe and has an energy 10·• eV higher fhan fhe lower level which is
two-fold degenerate. (a) Calculafe fhe infernal energy and enaropy or fhe sysaem if il is
prepared fo have two paraicl.. in fhe upper level. (b) If fhere is no change in fhe system
when it is brought into contact with a reservoir at a temperature T, calculale the tern·
perafure of fhe reservoir. (c) Wrife fhe parlifion funcfion for fhis sysfem. (d) Repeat
parts (a), (b), and (c) for the case fhaf fhe degenerofe level has an energy JO-• eV higher
than fhe nondegenerafe level.
11-30 (a) Derive Eqs. ( 11- 65) und (11-66) for a sysfem obeying M-8 sfafi Siics and in
which fhe energy levels are defermined by an exfensive paramefer X. (b) Show that fhe
expr<SSions for the infernal energy U and fhe infensive paramefer Y for fh is sysaem are
slill given by Eqs. (11-62) and (11-64).
I 1-31 (a) Using Eqs. (1 1-11) and (11-86) (Problem 11-22) for fhe fhermodynamic
probabilify of a macroSiafe of a sysfem of N panicles obeying M-B and classical sfafistics
respecfively, show fhat n,H, = N!!l,. (b) Use the resulf of para (a) fO show that the
 348 STATISTICAL THERMODYNAMICS

enlropies of the two systems are relaled by S61-D - s.

+ Nk 0 (1n N - I) and that the
Helmholtz functions are relaled by F>J·O - F0 + NkuT(In N - 1).
11-Jl Show that for a system of N particles obeying M·B or classical statistics the average
number of particles in the level j is given by
_ (alnZ\
N1 - -Nk8T ""T.;Jr• (11..$7)

11-33 (a) Derive an expression for the enthalpy of a system if the partition function
depends on X and T. (b) Derive an expression for the internal energy of a system if the
partition function depends on y and r.
11-34 Consider a system of N distinguishable particles distributed in two nondegenerate
levels separated by an energy < and in equilibrium with a reservoir at a temperature T.
Calculate (a) the partition function, (b) the fraction NJN and NJN of particles in each
state, (c) the internal energy U of the system, (d) the entropy S or the system , (e) the
specific heat capacity c. or the system. (f) Make sketches or N 1/N, N,/N, U, S, and c. as
a function of T .
11-35 Consider a system or N distinguishable particles each having a magnetic moment
p, distributed over two nondegenerate levels having energies lut'ol2 and - !'LJ2, when
the magnetic intensity is Jl'.. The particles in the upper level have their magnetic moments
antiparallel to the field and those in the lower level are aligned parallel to the field. The
system is prepared to have one-third of all the particles in the upper level and is isolated.
(a) Find the energy and the net magnetic moment of the system. (b) Calculate the change
of the energy and the change of the net magnetic moment of the isolated system when
the magnetic intensity is reversibly reduced to JI'J2 . . (c) Calculate the change in the net
magnetic moment of the system when the magnetic intensity is reversibly reduced to
JI'J2 but the energy of the system remains constant.
11-36 The system of the previous problem is in thermal equilibrium with a reservoir at a
temperature T. (a) Show that the partition function is given by
pJI'.
z- 2cosh 2kur·
(b) Derive expressions for U, E, S, F•, and M for this system and sketch curves of these
properties as a function of Tfor a fixed value of ;r.. (c) Use Eq. ( 11..$7) (Problem 11- 32)
to find how the number of particle$ in each level varies with Jl'0 and T.
11-37 The M-B statistics and the F-D statistics can be developed by calculating collision
probabilities for elastic collisions between two particles. If two particles-obeying M·B
statistics initially have energies <1 and • • and after the collision <a and '•• then
•• + •• - <•· - 6) + (<, + 6).
The number of collisions per unit time F is proportional to the probability /{<1) tha t each
initial state is occupied:
F.., - cf(• 1)/{<,).
Similarly, F~ , - </{<,)/(<,). In equilibrium, F 1•1 - F,_,.
(a) Show that/(<1) - e-'•/ >'1'
solves this equation. (b) Use similar reasoning to derive the F-D statistics. Here, how-
ever, the initial states must be filled and the final stales must be empty. Therefore the
 PROBLEMS 347

n umber of collisions per unit time is

F 1, 1 - </(<1)/(<J[I - /(<,))[1 - /(<,)).
Show that the equation Fu - F,,, can be solved by

I - /(<,) - ce•,t>'I'
/(•J
which yields an equation of the form of Eq. (11-40).
11-38 Another way to derive the diSiribution functions is to define a grand partition
fim<tlon '!II
If [n(p - •Jl
'!II -:~.exp ~j ·
and calculate values of /J :

(a) Show that

II- - __d__ In'!'/.
/ • - p)
leT

(b) Show that H - I gives the Fermi-Dirac distribution function. (c) Show that H - co
gives the Bose-Einstein distribution function.
 12
Applications of statistics to gases
12-1 THE MONATOMIC ID EAL GAS

12-2 THE DISTRIBUTION OF MOLECULAR VELOCITIES

12-3 EXPERIMENTAL VERIFICATION OF THE MAXWELL-BOLTZMANN

SPEED DISTRIBUTION. MOLECU LAR BEAMS .

12-4 IDEAL GAS IN A GRAVITATIONAL FIELD

12--5 THE PRINCIPLE OF EOUIPARTITION OF ENERGY

1 2~ THE QUANTIZED LINEAR OSCILLATOR

12-7 SPECIFIC HEAT CAPACITY OF A DIATOMIC GAS

 350 APPLICATIONS OF STATISTICS TO GASES 12-1

12- 1 THE MONATOMIC IDEAL GAS

We next apply the general relations derived in the preceding chapter to the special
case of a monatomic ideal gas consisting of N identical molecules each of mass m.
The molecules are indistinguishabl~, and as we shall show later, the average number
of molecules in each of the possible energy states, except at extremely low tem-
peratures where all real gases have liquefied, is extremely small. The proper
statistics is therefore the classical statistics (Section I 1- 11).
The first step is to calculate the partition function,

Z- Ig,exp=!!.
1 kT
This requires a knowledge of the energy <1 and the degeneracy g1 of each level.
We assume that the molecules do not interact except at the instant of a collision,
so that each is essentially an independent particle and has the same set of energy
levels as does a single particle in a box. It was shown earlier that the principles of
quantum mechanics lead to the result that the energy levels of such a particle are
given by Eq. (1 1-4).
n~h v- '
2 11

.,----
8m
(12- 1)

where n~ = n! + n: + n:, and n., n,, n, a re integers each of which can equal
I, 2, 3, ... , etc.
The degeneracy g1 of a level, or the number of energy states in the level, can
readily be calculated when the quantum numbers are small, as in the example in
Section 11- 2. In many instances, however, the energy levels of an assembly arc
very closely spaced relative to the value of the energy itself. We can then subdivide
the energy levels into groups of width A<1, including those levels with energies
between <1 and <1 + A <1• We refer to each of these groups as a macrolewl. Let
'*' represent the total number of possible stat~s in all energy levels up to and
including the energy <1• The number of possible states A '*' within the macrolevel
is equal to the number of states in all levels incl uded in the macrolevel. That is,
6 '*1 is the degeneracy of the mocrolevel, but it arises in part from the grouping
together of a large number of levels, while the numbers g1 are fixed by the natu re
of the assembly.
lmagiqe that the quantum numbers n,, n,, " • are marked off on three mutually
perpendicular axes, as suggested in the two-dimensional diagram of Fig. 12- 1.
Every triad of integral values of n., n,, n, determines a point in what can be called
"n·space," and each such point corresponds to a possible state, provided the
quantum numbers are positive. We can think of each point as located at the center
of a cubical cell, each of whose sides is of unit length and whose volume is there·
fore unity.
The quantum number n1 corresponds to a vector in n-space from the origin
to any point, since n; - n! + tt! + 11:. In a system of given volume, the energy
 12-1 THE MONATOMIC IDEAl GAS 351

..
Fig. 12-l Quantum slates in n-space.

depends only on n1, so that all states of equal energy lie on a spherical surface
of radius n1 with center at the origin. Since n., n., and n. are all positive, and
since there is one point per unit volume of n·space, the total number <:§1 of
possible states, in all levels up to and including the energy.,, is equal to the volume
of one octant of a sphere of radius n1. That is,

(12- 2)

The spherical surface will of course cut through some of the unit cells and it is
not certain whether a point representing an energy state lies inside or outside the
surface. H owever, when n1 is a large number, as is the case for the vast majority
of molecules of a gas at ordinary temperatures, the uncertainty becomes negligibly
small.
The number of states in the macrolevel between <1 and ., + I!J.<1, or the
degeneracy D. <:§1 of the macrolevel, is

D. <:§1 = !!. x 3n: D.n1 = !!. n~ D.n 1• (12-3)

6 2
G eometrically, this corresponds to the number of points in a thin spherical shell
of radius n1 and thickness D.n1• The degeneracy therefore increases with the square
of the quantum number n1, for equal values of D.n1•
The partition function Z for this system is written

Z- };D. <:§1 exp~ ,

1 kT
 352 APPUCATIONS OF STATISTICS TO GASES 12-1

and on i7serting the expressions for A~~ and <1 , we have

., }.;n~exp
Z =- ( - h'v-"'
--n~ ) An1• ( 12-4)
2 1 8mkT
This sum can be interpreted graphically as follows. Let the values of n1 be; marked
off on a horizontal axis, and for brevity represent the coefficient of An1 in Eq. (12-4)
by f(n1). At each value of n1, we construct a vertical line of lengthf(n1), as in Fig.
12-2. Each productf(n1) An1 then corresponds to the area of a rectangle such as
that shown shaded in Fig. 12- 2, and the value of Z corresponds to the sum of all
such areas over values of n1 from j .. I to j .. co, since there is no upper limit to
the permissible values of n1• To a sufficiently good approximation, this sum is
equal to ihe area under a conti nuous curve through the tops of the vertical lines,
between the limits of 0 and co, so

Z = !f.m
2 o
nl exp ( - h'v-•ta n:) dn
8mkT
1
• (12-5)

The value of the definite integral can be found from Table 12-1, and finally,

z = vC'"~k~.,.· (12~
The partition function therefore depends both on the temperature T and the
volume V, which corresponds to the general extensive variable X in Section 11- 15.
The Helmholtt function F is given by Eq. (1 1-63) as

F - - NkT(ln Z -InN + I),

Fig. Il-l The partition function Z is

equal to the total area under the step
fu nction, and is very nearly equal to the
area under the continuous curve.
 12- 1 THE MONATOMIC IDEAL GAS 363

Table 12-1 [(n) =f. ..x•e-..• dx.

n [(n) n f(n )
---
0 ~A
2 Q
I
20
I

~Jf.
I
2 3 2Qi

4 Jfi
8 ;;; 5
I
;;;
3
6 15Jf.,
i6 ;;; 7 ;;<
+..
If n is even,
i
_.,
+«>
xne-u' dx = 2f(n).

Jf n is odd,
f-.. x"t-oz' dx - 0.

and the pressure P, which corresponds to the intensive variable Y, is

P = Nkr( o In
oV
z) .T
(12-7)
Since by Eq. (12- 6),
lnZ = In V +~2 I n ( 2"""kT)
h2 •
( 12-8)
it follows that
(12-9)
Consequently
p = NkT = nRT, (12- 10)
v v
which is just the equation of state of an ideal gas as derived from kinetic theory.
The internal energy U is

U = NkT-. ( -
olnZ)
- = 3-Nk T = -3 nRT, (12-11)
oT I ' 2 2
which also agrees with the results of kinetic theory for a monatomic gas havi ng
three degrees of freedom.
 354 APPUCATIONS OF STATISnCS TO GASES 12- 2

The heat capacity at constant volume is

c.,= (8Tu)v
8
= ~Nk = ~nR
2 2 '
(12-12)

and the molal specific heat capacity is

c.= Cv = ~R. (12- 13)
n 2
The entropy is

S- !:!. + Nk(ln Z - In N + 1),

T
and after inserting the expressions for In Z and U, we have

S- Nkr~ +In V(2,mkT)"1. (12-14)

[j Nh' J
The principles of thermodynamics define only difftrmcts in entropy;
the expression for the entropy itself contains an undetermined constant. There
are no undete rmined constants in Eq. (12-14) and the methods of statistics there-
fore lead to an expression for the entropy itself.
Using Eq. (12-13), the molal specific entropy can be written

(2,mk}'1'
s =c. In T + Rln V + R [ In N'hl" + 25] · (12- IS)

This agrees with the thermodynamic expression for s in its dependence on V and
T, and contains no undetermined constants. Equation (12-15) is known as the
Sackur•-Tetrodet equation for the absolute entropy of a monatomic ideal gas.

12-2 THE DISTRIBUTION OF MOLECULAR VELOCITIES

In the chapters describing the kinetic theory of gases, a number of results were
obtained w~ich involved the average or root-mean-square speed of the molecules,
but at that lime we could say nothing as to how the molecular speeds were distri-
buted around these average values. (We use the term "speed" to mean the mag-
nitude of the velocity.) The methods of statistics, however, lead directly to the
expression for the occupation numbers of the energy levels a nd hence to the speed
distribution. An expression for the distribution was first worked out by Maxwell,
before the development of statistical methods, and later by Boltzmann and is
referred to as the Maxll't/1-Bo/tzmann distribution.
As in the previous section, we express the distribution in terms of the average
occupation number of a macrolevel, including an energy interval between t 1 and

• Ouo Sackur, German chemist (1880-1914).

t Hugo M. Tetrode, Dutch physicist (1895-1931).
 12-2 THE DISTR IB UTION OF MOLECULAR VELOCITIES 315

~, + A<1 . Let ,V represent the total num ber of molecules with energies up to and
including the energy '•· The average number of molecules included in the macro-
level, or the average occupation number of the macrolevel, is then ~.. The
quantities ~~ and Af/1 then correspond to the occupation number Fl, and de-
generacy g1 of a single energy level and both the M-B and classical distribution
functions can be written
N
~1 --Afl (-~,)
Z
1 exp -.
kT
(12-16)

Because we are interested in the distribution in spud rather than in enugy,

we express the degeneracy Afl1 in terms of the speed v, instead of the quantum
number n1• We have from Eqs. (12- 1) and (12-3),

n'h•v-11• 1 1
£,=~:::~2mvl.

A fl1 - in,An 1•

It follows from these equations that

m•v v•Av.
A ':I• -- 4 ..h. (12- 17)

For simplicity, we have dropped the subscript j from v, and written Afl. to
indicate that the degeneracy is expressed in terms of v. Finally, taking the expres-
sion for Z from Eq. (12-6), we have

~
•
=
4
':_(....!!!..\"'u•exp ( -
.,;.. 2kT}
mv•) Au.
2kT
( 12-18)

The quantity .,v. represents the average total number of molecules with all speeds
up to and including v, and A.,V. is the average number with speeds between v
and v + Av.
It is helpful to vis ualize the distribution in terms of"velocity space." Imagine
that at some instant a vector vis attached to each molecule representing its velocity
in magnitude and direction, and that these vectors are then transferred to a common
origin, resulting in a sort of spiny sea urchin. The velocity of each molecule is
represented by the point at the tip of the corresponding velocity vector. Figure
12-3 shows one octant of this velocity space. Geometrically speaking, the quantity
,v. represents the average total number of representative points within a sphere
of radius u, and ~. the number within a spherical shell of radius u and thick-
ness Au.
The coefficient of Au in Eq. (12-18), equal to the ratio A.,V./Au, depends only
on the magnitude of v, or on the speed. It is called the Maxwell-Boltzmann speed
distribution function and is plotted as a function ofv on Fig. 12-4. The number of
I
 12-2
3M APPLICATIQNS OF STATISTICS TO GASES

..

.,

Fig•.ll-3 Diagram of velocity spaoe.

Fig. Jl-4 Graph of Maxwell·Boltzmann

speed distri bution function.
 12-2 THE OISTAIBUTION OF MOLECUlAR VE LOCITIES 367

velocity vectors A.,V. terminating between v and v + l1v is represented in th is

graph by the area of a narrow vertical strip such as the shaded one shown, since
t he height o f the strip is t;..;VJt>v and iLS wid th is t.v. {Note carefully that the
ordinate of the speed distribution func tion does not represent t;..;V•. ) The d istri·
bution function is zero when v '"' 0, since then v• .. 0 and the exponential term
equals I . This means tha t no molecules {or very few molecules) a re at rest. The
function rises to a ma.x imum and then decreases because t he exponential term
decreases more rapidly than v' increases.
If velocity space is subdivided into spherical shells of equal t hickness, the
speed v,. a t which the distributio n function is a maximum is the ra dius of that
spherical shell which includes t he largest number of representative points. The
speed v,. is called the mo$1 probabl~ spud. To find iLS value, we take the first
derivative of the distribution function with respect to v and set it equal to zero.
Neglecting the constant terms in Eq. (12- 18), this procedure yields: J

.!...[•' exp ( -mv')]

dv 2kT
.. 0.
It is left as a problem to show that
•• = ,ftT<17iii. (12-19)
The d istribution function can now be expressed more compac tly in terms of

l1.AI".
- - .. -4N- v, exp ( --•) . ( 12- 20)
~v .J"v!. v!_
T he distribution function depends o n the temperature of the gas throug h the
quantity "•• which appears both in the exponentia l function and iLS coefficient.
Figure 12- 5 is a graph of the distribution function at three different temperatures.

Fig. 12-S Graph of M· B spud dislribution

(unction at 1hree different temperatures,
T0 > T2 > T1•
 368 APPLICATIONS OF STATISTICS TO GASES 12- 2

The most probable speed decreases as the temperature decreases and the "spread"
or the speeds becomes smaller. The areas under all three curves are equal , since
the area corresponds to the total number or molecules.
As explained in Section 9-3, the average or arithmetic mean speed is
I
D=NivAA"•.
Using Eq. (12-20) and approximating the sum by an integral, we have

ii= ,~ . J."'v•cxp(-,0~dv.
"'"""' • v,.-J
The definite integral, From Table 12-1, is o./2, so

v=2v,.=
.J7r ~.
~;-;;;
( 12- 21)

The root-mean-square speed is

v,m,- ..;=,} = 4 J."' v,exp (-v')

(~I v'AA".r= [ J;.v:,. v!. dv]'''.
0

The definite integral equals

3
f v>, so

( 12-22)

which agrees with Eq. (9- 19) obtained From kinetic theory. The method used here
is Far more general than that used to derive Eq. (9- 19). The method is applicable
to systems more complicated than an ideal gas by changing the dependence or
• 1 and g 1 on the velocity or the particles.
In summary, we have

The three speeds are shown in Fig. 12- 6. The relative magnitudes oF the three, at a
given temperature, arc
v. :u:v,m, • I :1. 128:1.224.
The quantity AA". represents the number or velocity vectors terminating in a
spherical shell in velocity space, or "volume" 4,v• Av, between o and o + Av.
The number or representative points per unit " volume" within the shell, or the
 12-2 THE DISTRIBUTION OF MOLECUlAR VELOCITIES

Fie. 12-6 Most probable (v.), arithmetic

mean (ii), and root·mean·square (v,., )
speeds.

"density" p. in velocity space, is

p - ~A". =
1 )•exp (- •'\
N(-.,foo,. (12-23)
• 4ml Ao u:: J'
The quantity p. is called the Maxwell-Boltzmann velocity distribution function. It
is a maximu m at the o rigin, where o = 0, and decreases exponentially with o' as
shown in Fig. 12- 7.
Note that although the density is a maximum at the origin, the spherical shell
containing the largest number of representative points is that of radius v... The
reason for this apparent discrepancy is that as we proceed outward from the origin,
the volumes of successive spherical shells of equal thickness ~u continually in·
crease, while the number of representative points per unit volume continually
decreases. The volume of the innermost shell (which is actually a small sphere of
radius Au) is essentially zero, so that although the density is a maximum for this
shell, the number of points within it is practically zero because its volume is so
small. Jn other words, practically none of the molecules is at rest. Beyond the

Fig. 12-7 Graph or Ma•well·

Boltzmann ~locity distribution
function.
 360 APPLICATIONS OF STATISTICS TO GASES 12-2

sphere of radius u,., the density decreases more rapidly than the shell volume in-
creases and the numbe r of points in a shell decreases.
The number of molecules f1.AI',., having specified values of all three
velocity compontnts corresponds, in Fig.' 12- 3, to the number of representative
poiniS within a small rectangular volume element in velocity space having sides
of length Au•. Au•.
and Au., and located at the point "·· u•• u,. The volume of the
element is Au. Au, Au, and the number of representati ve points within it is the
product of iiS volume and the density p, . Thus

I
= N( .;; "'"'
)'
exp
[-(u! +v!.u! + u:)J .1oa .6.u,. Av.,
since u1 -= v! + v! + v!.
The number of molecules having an x-,y-, or z-component of velocity in some
specified interval, regardless of the values of the other components, is represented
in Fig. 12-3 by the number of representative points in the thin slices perpendicular
to the velocity axes. (The diagram shows only the intersections of these slices
with planes perpendicular to the axes.) Thus to find the number of molecules
f1.AI'. with velocity components between o, and u, + Av,, we sum A%, , • over
all vaiues of u, and u,. When the sum is replaced with an integral, we h~v"e'

f1.AI',, = NCd UJ[[exp c~=) du,[exp (~~:) du,] exp (~~:)Au•.

Each of the integrals, from Table 12- 1, equals..{; u,., and therefore

f1.AI',,
- - = N -I- e x p - (-u!) (12-24)
Au. ~; v,. v~ '
with similar expressions for u, and u,. These are the Maxwell-Boltzmann distri-
bution functions for ont component of velocity, and that for the x-component is
plotted in Fig. 12-8. The slice in Fig. 12- 8 containing the largest number o f
representative points is therefore the one at u, - 0, and the most probable velocity
component along any axis is zero.
The distribution represented by Eq. (12-24) and Fig. 12-8 is known as a
gaussian• distribution and is typical of many sorts of random distributions, not
just that of molecular velocity components. This is to be expected, since the treat-
ment that led to Eq. (12- 24) is so very general.
We can now show that it is appropriate to use the classical distribution func-
tion to describe an ideal monatomic gas. II will be recalled that the Bose-Einstein
and Fermi-Dirac distribution function both reduce to the classical distribution

• J. Carl F. Gauss, German mathematician {1777-ISH).

 12-2 THE DISTRIBUTION OF MOLECULAR VELOCITIES 311

Fig. ll-8 Maxwell-Boltzmann veloc-

iry disrriburion funcrion for a single
componenl or velociry.

func lio n, provided Ihe occupation numbers A%1 are much smaller than the number
of sta tes tJ.'§1 in the macrolevel j. In other words, the classical distri bution func-
tion is applicable p rovided t:..A',!M/1 « I. According to Eq. (12- 16), t he general
expression for tJ..A',/M/1 in this case is

!J..A',
- - = -Ne x (p
-•,)
-
tJ.~1 Z kT '
and for an ideal gas,
21Tmk'!}311
Z=V-
( h-,-, .
Therefore
tJ.%1 N( 2,.mk1)..." ( -•1)
tJ. <§ =
1
Y
-h-,--1 exp kT ·
Lei us rake u an example helium gas at srandard condirions. I n a Maxweii-
Bollzmann velociry disrriburion, rhe energies ., arc grouped around the mean value
3kT{l. Then •,/kT is of rhe order of uniry and so is exp (- •,/k1). The number of
mo lecules per unit volume, N/V, is aboul 3 x rou molecules m-•
and for helium,
m - 6.7 X ro-n kg. Jnserring lhe values or"· k, m, and Tin the preceding equation,
we get
t..IY, •
A'§, ::. 4 X 10",

which is certai nly much less rhan uniry. (Only about four slates in a million arc
occupied!) However, as rhc lemperaturc is lowered, rhc value of t..IY' 1/A~1 increases,
and provided the gas can be cooled to very low temperalures withoul condensing,
the classicalsrarisrics may cease to be applicable. Conversely, condensation may well
be a djusted just when the classical statistics cease to be applicable, and this reftects
the essenlially quantum-mechanical nature of liquid helium.
 312 APPLICATIONS OF STATISTICS TO GASES 12-3

12-3 EXPERIMENTAL VERIFICATION OF T HE M AXWELL-BOLTZMANN

SPEED DI STRIBUTION. MOLECULAR BEAMS
An important technique in atomic physics is the production of a collimated beam
of neutral particles in a s<Kalled moltcular Mam. A beam of chargd particles,
electrons or ions, can be accelerated and decelerated by an electric field, and guided
and focused by either an electric o r a magnetic field. These methods cannot be
used if the particles are uncharged. Molecular beams can be produced by allowing
molecules of a gas to escape from a small opening in the walls of a container into a
region in which the pressure is kept low by continuous pumping. A series of •
baffles, as in Fig. 12- 9, limits the beam to a small cross section. Since one often
wishes to work with molecules of a material such as silver, which is a solid at room
temperature, the temperature in the container must be great enough to produce
a sufficiently high vapor pressure. Hence the container is often a small electric
furnace or oven.
We have shown in Section 9-3 that the number of molecules with speed v,
striking ~e surface of a container per unit area and per unit time, is
h
~vAn. (12-25)
F
d
where An. is the number o f molecules per unit volume with speed v. d
If the molecules have a Maxwell-Boltzmann speed distribution, the number 6
per unit volume with a speed vis given by Eq. ( 12-18) tl

An. = ;,( 2:Trv• exp ( ~:;') Av.

If there is a hole in a wall of the oven, small enough so that leakage through
the hole does not appreciably affect the equilibrium state of the gas in the oven,
Eq. (12-25) gives the number with speed v escaping through the hole, per unit
area and per unit time. We wish to compute the rms speed of those that escape.
Following the standard method, the mean-square speed of the escaping molecules
Is found by multiplying by v' the number that escapes with speed v, integrating
over all values of v, and dividing by the total number. The rms speed is the square

o ... Baines

Fig. 12-9 Production or a beam or

neutral particles.
 12-3 VERIFICATION OF THE MAXWELL·BOLlZMANN SPEED DISTRIBUTION 3e3

root of the result. It is left as a problem to show that

·=· =ff!..
The rms speed of t he molecules in the oven is
(12-26)

v, .., =fff..
so that those escaping have a somewha t higher speed t han those in the oven .
The distribution in direction of the molecules escaping thro ug h the hole is
given by Eq. (9--14):
t.ci>m=..!_ iin cos O.
f. w 41T
That is, the number per unit solid angle in the emerging beam is a ma~ imum in the
direction of the normal to the pla ne of the opening and decreases to zero in the
tangential direction.
Direct measurements of the distribution of velocities in a molecular beam
have been made by a number of methods. Figure 12- 10 is a diagram of the ap-
paratus used by Zartman and Ko in 1930-1 934, a modification of a technique
developed by Stern in 1920. In Fig. 12-10, 0 is an oven and S1 a nd S1 are slits
defining a molecular beam; C is a cylinder that can be rotated at approximately
6000 rpm a bout the a~is A. If the cylinder is at rest, the mo lecular beam e nters
the cylinder through a slit s. and stri kes a curved glass plate G. The molecules

--:--s, I

--i- -s,
C=Jo
Fig. IZ- 10 Apparatus used by
Zartman and Ko in studying
distribution of velocities.
 364 APPLICATIONS OF STATISTICS TO GASES 12-3

stick to the glass plate, a nd the number arriving at a ny portion can be determined
by removing the plate and measuring with a recording microphot6meter the
darkening that has resulted.
Now suppose the cylinder is rotated. Mo lecules can enter it only during the
short t ime intervals during which the slit S crosses the molecular beam. I f the
rotation is clockwise, as indicated, the glass plate moves toward t he right while
t he molecules cross the diameter of the cylinder. They therefore strike the plate
at the lefi of the point of impact when the cylinder is at rest , and th e more slowly
t hey travel, the farther to the lefi is this point of impact. The blackening of the
plate is therefore a measure of the "velocity spectrum" of the molecular beam.

I
I
I
I
I
I
I

,.;.::.::.·..::.·..:_·..:._·.::._·.::....1 !o
~-~-~- ~ ~~~~~- :
"" . . . . . . . . . . . . . . . . . -io·
I
I
', I
.... ~ o·

Fi1:. 12-11 Schematic diagram or apparatus or Estermann, Simpson,

and Stem.

A more precise experiment, making use of the free fall of the molecules in a
beam, was performed by Estermann, Simpson, and Stern in 1947. A simplified
diagram of the apparatus is given in Fig. 12- 11. A molecular beam of cesium
emerges from the oven slot 0 , passes through th e collimating slitS, and impinges
on a hot tu ngsten wire D . The pressure o f the residual gas in the apparatus is of
the order of 10~ Torr. Both the sli ts and t he detect ing wire are horizontal. The
cesium atoms striking the tungsten wire become ionized , reevaporate, and a re
collected by a negatively charged cylinder surrounding the wire but not shown in d
the diagram. The ion current to the collecting cylinder then gives d irectly the r
number of cesium a toms impinging on the wire per second .
In the absence of a gravitational field, only those atoms emerging in a hori- a
zontal direction would pass through the slitS, and they would all strike the collector t
in the positi on D regardless of their velocities. Ac tually, the path of each atom is )
a parabola, and an atom emerging from the slit 0 in a horizontal direction, as n
 12-3 VER IFICATION OF THE MAXWELL-BOLTZMANN SPEED DISTRIBUTION 365

indicated by t he dot and dash line, (with the vertical scale greatly exaggera ted)
would not pass through t he slit S. The dashed line and the dotted line represent
the t rajectories of two atoms that can pass t hrough t he slitS, the velocity a long the
dashed trajec tory being greater than t ha t along the dotted one. Hence as the
detector is moved down from the pos ition D, those ato ms with velocities corre-
sponding to the dashed trajectories will be collected at D' , those wi th t he slower
velocity corresponding to the dolled trajectory will be collected at D•, etc. Measure-
men t of the ion c urrent as a fu nction of the vertical height o f the collector then give•
the velocity distribution.

Fig. 12- 12 Experimental verification of the

Maxwell-Boltzmann speed distribution fu nction.
This is Fig. 1 from R. C. Miller and P. Kusch,
"Velocity Distri bution in Potassium and
Thallium Atomic Beams," Plrysical Rtvltw 99
(1955): 1314. Reprinted by permission.

In 1955 Miller a nd Kusch reported a still mo re precise measu re ment of the

d istribu tion of velocities in a beam of thallium atoms. Their data are shown in
Fig. 12- 12. The oven, which was controlled to 0.2S' C, was made from copper
to insure a uniform temperature distribution. The t hallium a toms passed throug h
a slit whose d imension para llel to the bea m was 0.003 em to avoid scattering in
the neighborh ood of the slit. The detecto r was simila r to the previous experilnent.
As the a toms came out of t he slit they had to pass through one of 702 helical slits
milled a long the surface of cylinder 20 em in diameter, 25.4 e m in length. Each
 3U APPUCATIONS OF STATISTICS TO GASES 12-4

slit was 0.04 em wide and 0.318 em deep. As the cylinder was rotated, only those
atoms having an appropriate velocity would pass through the slit without being
scattered. With these precautions Miller and Kusch were able to show that the
velocity distri bution of the thalliu m atoms agreed with the Maxwell-Bo ltzmann
velocity distribution to withi n I % for 0.2 < x < 1.8, where x - vfv~. This
agreement is seen on Fig. 12- 12 where the points are the data fo r two d ifferent 1
experiments and the solid line is the theoretical curve computed from the Maxwell- e
Boltzmann speed d istribution.

12-4 IDEAL GAS IN A GRAVITATIONAL FIELD

In the precJding sections, the energy of a gas molecule was considered to be wholl,
kinetic; that is, a ny gravitational potential energy of the molecule was ignored.
We now take this potential energy into account, so that the gas serves as an example n
of a multi variable system.
Let us take as a system an ideal gas in a vertical cylinder of c ross-sectional area
A, as in Fig. 12-13. The lower end of the cylinder is fixed and the upper end is

Fi&. ll-13 An ideal gas in a

cylinder in a gravitational field.

provided with a movable piston. If the piston is a t a height L above t he bottom

of the cylinder, the vol ume V occupied by the gas is V = AL. The origin of space
coordina tes is at the bottom of t he cylinder, with they-axis verlically upwa rd.
The system is in a uniform gravitational field of intensity g, direcled vertically
downward ; but the value ofg can be changed by, say, moving the system to another
location whereg has a different value. The temperatu re Tis assumed to be uniform.
The gas is therefore a multivaria ble system, described by three independent
variables T, L, and g, and it has a gravitational potential energy E. as well as an
internal energy U. The appropriate energy function is therefore the total energy
 IDEAL GAS IN A GRAVITATIONAL FIELD 317

given by
E=U+E,,.
d from Eq. (7-31),
TdS ~ dE+ Y1 dX,- X1 dY,.
•e extensive variable X, is the length L , and the intensive variable Y, is the
avitational field intensity g. Let us represent the variable Y, by n, and the
Iiable X, by r. Then
T dS = dE + n dL- r dg. (12- 27)
·e now use the methods of statistics to find the quantities n and r. The first
:p is to determine the partition function Z .
A molecule whose vertical coordinate is y has .a gravitational potential enqrgy
V in addition to its kinetic energy mu1/ 2, and its total energy < is
• - mv'/2 + mgy.
An energy interval between • and < + tu includes a kinetic energy interval
>rresponding to speeds between v and v + Au, and a potential energy interval
>rresponding to elevations between y andy + t:.y. The degeneracy t:.!l. of the
>eed interval, since V- AL, is given by Eq. (12-17),

t:.!l _ 4"m'AL ' t:. (12-28)

•- h' v v.
The potential energy is not quantized ; a molecule may have any arbitrary
levation y and any potential energy mgy. The distribution in potential energy
' given by the same expression as that for quantized levels, however, if we set
Je degeneracy t:.!l, of the potential energy interval equal to t:.yfL:

t:.f§ ~ t:.y (12-29)

• L
For any one of the possible kinetic energy states, a molecule can have any
me of the possible potential energy states. The total number of possible states
l':l in the energy interval is therefore the product of t:.!l, and t:.<§,:
!:.':1 ~ t:.!l,t:.':l,.
The partition function Z is

z-!t:.'lexp(;;)
- [ }:!:.':1, exp ( ~:;')] [!t:. ':1, exp ( - km:y)J (12-30)

lf we designate the sums by Z, and Z,, respectively, then

z- z.z,. In Z - In Z, + In Z,.
 368 APPLICATIONS OF STATISTICS TO GASES · 12-4

T he first sum in Eq. (12-30) is to be evaluated over all values of u from 0 to

co, and the second over all values of y from 0 to L . When the expressions for
6 '!1, and 6 -:?. are inserted, and the sums replaced with integrals, we find
2rrmkT\'11
z. = AL ( - -,-, ,
11
{12- 31)

kT
Z, - - -
mgL
[r - exp (- mgL)] .
kT
(12-32)
Therefore

In Z- ~In T- In g + In [r - exp ( -~•/L)J+ constant. {12-33)

The function F* is given by Eq. {11-75),

F* = -NkT(In Z - In N + 1),
and F* is a function of N, T, g, and L.
If N is constant,
n =-(oF") = Nkr( o ln Z) •
oL .... oL ....
and
r= (oF*)
or ... r.
_ -Nkr(oln
or
z)T.r..
On carrying out the differentiations, we find
n_ Nmg
{12-34)
- exp (mgLfkT) - 1 '
r= Nk T _ NmL (12- 35)
g exp (mgLfk T) - 1
Thus the system has two equations of stale, one expressing n as a funct ion of
T, L, and g, and the other expressing r as a function of these variables.
The physical significance of r can be seen as follows. The gravitational
potential energy £ 0 is
£ 1, - Y,X, = gr,
and hence

r=~.
g
Thus r is the potential energy, per unit field intensity. The potential energy is
therefore I
NmgL
E0 = gr = NkT -
exp (mgLfkT) - I
. (12-36)
 2-4 ID EAL GAS IN A GRAVITATIONAL FIELD 369

T he total energy E is

E ~ NkT' ( o In Z) - ~ NkT - NmgL ' (12- 37)

oT L., 2 exp (mgLfk T) - I
nd since U = E- £ 0 , it follows that
3
U = 2NkT .

fence the internal energy is the same as in the absence of a gravitational field and
epends only on the tempera ture.
T he entropy can be calculated from

S - §. + Nk(ln Z - InN + 1).

T
We next calculate the pressure P as a fu nction of elevation. The number of
1olecules A%, in a macrolevel betweeny andy+ fly is, from Eq. (12- 16),

fl.A", - -N fl <§, exp (-

-mgy)
- . (12-38)
Z, kT
'he volume of a thin cross section is A fly, so the number of molecules per unit
olume a t a height y is
fl.AI'.
"r z: A 6y:
'rom the ideal gas law, the pressure P, a t this height is
P, = n,kT.
It follows from the preceding th ree equations, after inserting the expressions
or fl'd, and Z,. that
P _ Nmg exp ( -mgyfk T )
• A I - exp (-mgL/kT)
It the bottom of the container, y - 0, and the pressure P 0 is
Nmg
Po = - - ---~-~::-=
A I - exp (- mgLfkT)
' he pressure P, can therefore be written more compactly as

- mgy)
P, • P 0 exp ( ~, (12-39)

.nd the pressure decreases exponentially wi th elevation. Equation (12- 39) is

.nown as the baromnric ~qua/ion or the /all' of almosphrrn. It can also be derived
rom the principles of hydrostatics a nd the equation of state of a n ideal gas.
 370 APPUCATIONS Of STATISTICS TO GASES 12-5

At the top of the container, y • L and

PL = Nmg n
A exp (mgL/kT) - I = A·
Therefor9
n - P~. (12-40)
a nd the quantity n is the force exerted agai nst the piston at the top of the con-
tainer. The work when the piston is displaced upward by an·amount dL is

dW - D dL- PLA dL- PLdV,

and the product n dL is the work when the gas expands.
In 1909 Perrin• used Eq. (12-39) in one of the earliest precision determinations
of Avogadro's number NA- Instead of gas molecules, he utilized particles of micro-
scopic size suspended in a liquid of slightly smaller density, thus reducing the
effective value of"g". The num ber of particles at different levels was counted wit h
a microscope.
If ~. and ~. arc the average numbers at heights y 1 a ndy1 , then

A%, = exp [ - mg(y, - Ya)J. (12-41)

A% 1 kT
All of the quantities in this equation can be measured experimentally with the
exception of the Boltzmann constant k , so that the equation can be solved fork.
Then N 4 can be found, since k equals the universal gas constant R divided by N 4 ,
and R is known from other experiments. Perrin concluded that the value of N 4
lay between 6.S x JO'" and 7.2 x 10'", compared with the present best experi-
mental value of 6.022 x 10" molecules kilomoJe- •.

12-5 THE PRINCIPLE OF EQUIPARTITION OF ENERGY

It will be recalled that the principle of equipartition of energy was introduced in
Section 9-6 merely as an inference lhat might be drawn from some of the results
of the kinetic theory of an ideal gas. We now show how this principle follows from
the M·B or classical distribution function and what its limitations are.
The energy of a particle is in general a function of a number of different
parameters. These migh t be the velocity components, the vertical elevation of the
particles in a gravitational field, the angle that a molecular dipole makes with an
electric field, and so on. Each of these parameters is called a degru offrudom.
Let z represent any such para meter and •(z) the energy associated with that
parameter. If the energy can be expressed as a continuous function of the param-
eter, as in the preceding sections, the M· B and classical distribution function lead

• Jean Perrin, French physicist (1870-1942).

 12-5 THE PRINCIPLE OF EOUIPARTITION OF ENERGY 371

to the result that the average number of particles within a range t:.z of the parameter
is given by an expression of the form

tvr', = A exp [ -k•~z)J 6 z,

where A is a constant independent of z. As examples, see Eq. ( 12- 24) for the case.
in which z represents one of the rectangular components of the velocity, or Eq.
(12-38) in which z represents the verrical coordinate y.
When the sum is replaced with an integral, the total number of particles, N,
is given by

N- AJexp [ -k•~•>J dz,

the limits of integration being over all values of z.
The total energy £(z) associated with the parameter z is

E(z) = J J
<(z) d%, = A <(z) exp [- '~i] dz.
The mean energy t(z) of a single particle is

i(z)- E(z) .
N
Now if the energy <(z) is a quadratic function of z, that is, if it has the form
<(z) - az•, where a is a constant, and if the limits of z are from 0 to oo, or from
- oo to +oo, then from Table 12- 1,

l(z) =
Jaz• exp ( -az1/ kT) d z
1
- - kT. (12-42)
J 1
exp ( -az fk T) dz 2

That is, for every degree of freedom for which the conditions above are fulfilled,
the mean energy per particle, in an assembly in equilibrium at the temperature T,
is kT/2. This is the general statement of the equiparrition principle.
The conditions above are fulfilled for the translational velocity components
v,, v,: and v,, since the energy associated with each is mo!/2, mv!/2, or mo!/2
and the range of each is from - oo to + oo. They are also fulfilled for the displace-
ment x of a harmonic oscillator, since the potential energy associated with x is
Kx1f2, K being the force constant.
The conditions are not fulfilled for the vertical coordinate y of a gas in a
gravitational field, where the potential energy is mgy; the mean gravitational
potential energy is not kT/2. Neither are they fu lfilled for the energy associated
with molecular rotation, vibration, and electronic excitation, because of the
 372 APPUCATIONS OF STATISTICS TO GASES 12-6

quantized character of these energies, which can take on only certain discrete
values and cannot be expressed as a continuous function of some coordinate. The
energy associated with them is not a simple linear function of the temperature.

12-41 THE QUANTIZED LINEAR OSCILLATOR

We consider next an assembly of N identical linear oscillators, assumed distin-
guishable so that we can use Maxwell-Boltzmann statistics. The properties of such
an assembly form the basis of the theo ry of the specific heat capacity of polyatomic
gases and of solids.
A linear oscillator is a particle constrained to move along a straight line and
acted on by a restoring force F ~ - Kx, proportional to its displacement x fro m
some fixed p oint and oppositely directed. The equation of motion of the particle is

d'x
F = m dt' ~ -Kx,

where m is the mass of the particle. If displaced from its equilibrium position and
released, the particle oscillates with simple harmonic motion of frequ ency •· given
by

., =.!. JK[;,.
2,.
The frequency depends only o n K and m, and is independent of the amplitude x,..
The energy • of the oscillator is the sum o f its kinetic energy mv'f2 and its
potential energy Kx'/2. Since the total energy is constant, and the kinetic energy
is zero when the displacement has its maximum value x,., the potential energy at
this displacement is equal to the total energy • and hence

,.! Kx!..
2
Thus the total energy is proportional to the squau of the amplitude, x,..
If the oscillators were compl~tely independent, there could be no interchange
of energy bftween them, and any given microstate of t he assembly would continue
indefinitely. We therefore assume that the interactions between the particles are
large enough so that there can be sufficient exchanges of energy fo r the assembly
to assome all possible microstates consistent with a given total energy, but small
enough so that each particle can oscillate nearly independently of the others.
In classical mechanics, a pa rticle can oscillate with any amplitude and energy.
The principles of quantum mechanics, however, restrict t he energy to some one
of the set of values

(12-43)
 !-6 THE QUANTIZED LINEAR OSCILLATOR 373

=
•here n1 0, I, 2, ... , and h is Planck's constant. An unexpected result is that
1e oscillator can never be in a state of zero energy, but that in the lowest level the
nergy ish•f2, in the next level it is Jh•/2, and so on. The levels are nondegenerate ;
:>ere is only one energy state in each level; and g1 - I in each level.
The quantum condition that the entrgy can have only some one of the set of
a lues [(n1+ l /2)h•J is equivalent to the condition that the amplitudt can have only
orne one of the set of values such that

x:O = (n1 + ~) ; JI/Km.

U sing Eq. (12-43), the partition function of the assembly is

Z- !exp (-•
I kT
1
) - ! exp
I
[-(n + !) .!!!..].
1
kT2
fo evaluate the sum, let z - h•/kT for brevity. Writing out the first few terms,
ve have
z ~ exp ( - n + exp ( - ¥) + exp ( - ¥) + ...
- exp ( - n{l + exp ( -z) + [exp ( -z)J' + .. ·}.
rhe sum in the preceding equation has the form of the infinite geometric series
I+ P + p' + .. · ,
.vhich equals 1/ (1 - p) as is readily verified by expanding the product ( I - p ) X
(I + p + p' + · · ·). Therefore
Z ~ exp ( - !)
1
2 I - exp ( -z)'
or
Z = exp ( -h•/2kT)
(12-44)
I - exp ( -h•/kT)
The temperature at which kT- /., is called the characteristic temperature of
the assembly and is represented by 0. Thus

kO = ho, or 0=!!!. (12-45)

k
It follows that
,. 0
kT ~r'
and in terms of 0 the partition function is

z= e xp ( -0/2T)
(12-46)
I - exp (-Of T)
 374 APPLICATIONS OF STATISTICS TO GASES 12-6

The value of the partition function at any temperature therefore depends, fo r

a given assem bly, o n the ratio of the actual temperature T to the characteristic
temperature 6, which thus provides a reference temperature for the assembly.
The greater t he natural frequency • of the oscillarors, the higher the characreristic
temperature. Thus if the natural freq uency is of the order of frequencies in the
infrared region of the electromagnetic spectrum, say 10,. Hz, • then

O ,. !!! = 6.62 X 1()"',. J s X IO'' s-• == SOO K.

k J.38 x 10-" J K- '
An actual tempe ratu re T of SO K is then approximately equal to 6/ 10, and a
tempera ture of 5000 K is approximately equal to 10 0.
T he average fractional n umber of oscillators in the jth energy level, from Eqs.
( 12-16) and (12-43) is

n
IV 1 I ( <1 ) l
( (n, + l)l••)
2
N - Z exp - k r •zexp - kT ·

Subsrituting Eq. (12-46) fo r Z a nd Eq. (12-45) for 0,

~ ~ [r- exp ( ~ ) Jexp ( - n ~

8
1 ) · ( 12-47)

At any temperature T, the occupation number decreases exponentially with the

q uantum number n1, and decreases more rapid ly, the lower the rem peratu re.
At the temperature at which T • 8,

(8/n• l,
and
!(
7:/ ~ 0.632 exp ( -n 1).

Thus for the four lowest energy levels, in which "J - 0, I, 2, and 3, we have
Flo R, Fl, R,
li- o.632, li - o.m, li - o.oss. li - o.on.
Abour 63% of the oscillarors are in the lowest e nergy level, abour 23% in the next Tl
level, ere. Togerher, rhe four lowest levels aecounr for about 98% of the oscillators. ter
It is left to the reader to show tha t when T • 0/2,

Ro R, Fl, lil,
li - o.86s. li - 0.111, li - o.o16, li = o.oo2.

• Heinrich R. Hertz, German physicist (1857-1 894).

 12-6 THE QUANTIZED LINEAR OSCI LLATOR 3711

Atrhis temperarure, abour87" of the oscillaton are in the lowest level, about 12"
in rhe nexrlevel, ere. and almost allrhe parricles are in the firsr four levels.
At a remperature T - 28,

n. -
N 0.394,
n, -
N 0.239,
ll.
N - o.I4S,
n, -
N o.o88.

The first fou r levels then accounr for only about 86% of the oscillaron, the remainder
being disrribured among rhe higher energy levels.
The lengrhs or the verrical lines in Fig. 12-14 represenr rhe average fractional
occuparion numbers ar rhe remperarures T - 8/2, T - 8, and T - 28.

1.0

0.8

lllli
fl, 06
N
0.4

0.2

"J • 0 I l .1 0 I 2 J 0 t 2 l
T• ~ T•D T• 20

Fig. 12- 14 The dependence on 8/T of the average

fracrional occuparion number of the first four ltvels
of a linear oscillator.

The total energy of the assembly, which in this case is its internal energy U, is

U- NkT'dln z
dT

- Nk8[ I
exp (8/T) - I
n.
+ iJ (12-48)

Thus for a given assembly of linear oscillators the internal energy is a function of
temperature only. The heat capaciry Cy of the assembly is

Cy = dU
dT
8)' exp (8fT)
- Nk ( - ~--~~~~2 (12-49)
T (exp (8fT) - 1] •
 378 APPLICATIONS OF STATISTICS TO GASES 12-7

Fia. 12-15 The internal energy

and heat capacity of an assembly
of linear oscillato11.

T he c urves in Fig. 12-1 S are graphs of the inter nal energy U and of the heat
capacity Cv (both divided by Nk) as functio ns of TfO. The ordinate of the latter
is proportional to the slope of the former.
As T approaches 0 K, very nea rly a ll of the oscillators are in their lowest energy
level with energy h•/2 and the total energy U approaches the zeropoint energy
Nh•/2, or, U/Nk - O.S. T he internal energy changes only slightly with changing
temperature and the heat capacity approaches zero. T he entropy of an assembly
of linearJscillators also approaches zero as T approaches zero.
Whe T » 0, 0/T« I, exp (0/T)- I ~ 0/ T, the term 1/2 is negligible com·
pared with T/0, and U approaches NkT. The mean energy per particle, U/N,
approaches kT which is the val ue predicted by equipartitio n fo r an oscillator with
t wo degrees of freedom (its position and its velocity). The internal energy increases
nearly linearly with temperature and Cv approaches the constant val ue Nk.

12-7 SPECI FI C H EA T CAPA CITY O F A DIATO MIC GA S

I t was shown in Sectio n 12- 1 how the equation of state of a monatomic ideal gas,
and its energy equation, could be derived by the methods of statistical thermo-
dynamics. Consider next a gas whose molecules are polyaromic. If the energy
of a molecule does not depend on the space coordinates x, y, and z of its center of
mass, and if there is no mutual potential energy between molecules, the partition
function will be directly proportiona l to the volume V, as in Eq. (12-6) for a
monatomic gas. The Helmholtz function F = -NkT(In Z- InN+ I) then
has the same dependence on Vas for a monatomic gas and the gas has the same
~quation of state, P V - nRT.
 12-7 SPECIF IC HEAT CAPACITY OF A DIATOM IC GAS 377

The specific .heat capacity, however, will differ from that of a monatomic gas
because a polyatomic molecule can have an "internal energy" of its own, made up
of energy of rotation, vibration, a nd electronic excitation.
According to the classical equipartition principle, each degree of freedom
associated with rotation and vibration shares equally with the three translational
degrees of freedom, the mean energy of each being kT/2. The molal specific heat
capacity at constant volume should equal R/2 for each degree of freedom and for
a molecule with f degrees of freedom we should have c. - JR/2, which should be
constant independent of temperature.
This prediction is in good agreement with experiment for monatomic gases,
for which there are three translational degrees of freedom only and for which c. is
ve ry nearly equal to 3R/2. At roo m temperature, however, the heat capacities of
diatomic gases are nearly equal to SR/2, as if the molecules had an additional two
degrees of freedom. Furthermore, the heat capacities are not constant, but vary
with temperature and do not correspond to integral values off
A diatomic molecule can be considered to have the dumbbell-like structure of
Fig. 9- 5. In addition to the kinetic energy of translation of its center of mass, it
may have energy of rotation about its center of mass and, since it is not a completely
rigid structure, its atoms may oscillate along the line j oining them. The rotational
and vibrational energy a re both quantized; and with each form of energy, as for
an harmonic oscillator, there can be associated a characteristic temperature, 0,0 ,
for rotation and 8, 1" for vibration. The extent to which the rotational and vibra-
tional energy levels are populated is determined by the ratio of the actual tem-
perature T to the corresponding characteristic temperature. That is, the internal
energies of rotation and vibration , and the corresponding specific heat capacities
c,., and c.,. are funct ions of the ra tios T/0,., and T/8,.11, . We shall not give the .
precise form of this de pendence, but simply state that the graphs of the specific
heat capacities c,...and c.,. c.
have the same general form as the graph of fo r an
harmonic oscillator shown in Fig. 12-15. At very low temperatures, both heat
capacities approach zero; at temperatures large compared to the characteristic
temperatures, both approach the classical value Nk. Thus at sufficiently high
temperatures the corresponding molal heat capacities app roach the classical value
R , as for a particle with two degrees of freedom.
What constitutes a "sufficiently" high temperature? This depends on the
characteristic te mperatures 8,., and 8,.11, . Table 12-2 lists some values of 8, 0,.
This temperature is inversely proportional to the moment of inertia of the molecule:
the greate r the moment of inertia, the lower the value of 8,.,. The highest value,
about 86 K, is that for hydrogen, H,, since its moment of inertia is smaller than
for any other diatomic molecule. Molecules with one hydrogen a tom form another
group wi th values of 0,.,
of approximately 20 K. For all others, the characteristic
temperature is of the order of a few degrees or less. Thus "room temperature,"
say 300 K, is much greater than the cha racteristic temperature for rotation, and
the molal specific heat capacity for rotatio n approaches the value R.
 371 APPLICATIONS OF STATISTICS TO GASES 12-7

Table Il- l Characteristic temperatures

for rotation and vibration of diatomic
molecules
Substance s,.,. (K) 8, 1., (K)

Ha ss.s 6140

OH 27.S S360

HCI JS.3 4300

CH 2fJ.7 4100

co 2.77 3120

NO 2.47 2740

Oa 2.09 2260

Cia 0.347 810

Bra 0.117 470

Naa 0.224 230

x. 0.081 140

Table 12-2 also lists the characteristic temperatures8,,. for the same molecules.
These are all very much higher than the characteristic temperatures for rotation,
which means that a t room temperature, whe re T « 8,,•• practically all molecules
are in their lowest vibrational energy level and the specific heat capacity for vibra-
tion is practically zero. Only at much highe r temperatures do the higher vibrational
energy levels begin to be populated.
T hus at room temperature the specific heat capacities of most diatomic
molecules have a contribution 3R/2 for translation, plus R for rotation, making
a total of SR/2 as is ac tually observed.
Figure 12-16 is a graph of experimental values of c./R for hydrogen, plotted
as a function of temperature. (Hydrogen is the only diatomic gas that remains a
gas down to low temperatures, of the order of 25 K.) At very low temperatures;
c.{R is equal to 3/2, the value for a monatomic gas. As the temperature is increased,
c. increases, and over a considerable range near room temperature c.JR is about
S/2, which is the value (aocording to equipartition) if two degrees of freedom of
rotation or vibration, but not both, are added to the translational degrees of
freedom. Only at very high temperature does cJR approach 7{2, the value pre- n
dieted by equipartition. to
 12-7 SPECIFIC HEAT CAPACITY OF A OIATOMIC GAS 371

We can now understand in a general way the features of this graph. The
characteristic temperatures for ro tation and vibration, for hyd rogen, are 8,.., =
85.5 K and e.,. - 6140 K. Below about SO K, the temperature T is very much
less than either characteristic temperature, and practically all molecules remain
in thei r lowest energy states of rotation and vibration. The specific heat capacity
is therefore the same as that of a monatomic gas, 3Rf2.
In the range from about SO K to about 250 K, the temperature T is of the
order of magnitude of 6,.,
and the rotational states of higher energy begin to be
populated. Above about 250 K , the molecules behave like classical rotators and
make a contribution R to the specific heat capacity, which in this range equals
5Rf2. Starting at about 500 K, some molecules move to states of higher vibrational
e.nergy and c. approaches the limiting classical value of 7R/2 .

•
,."
3
v
c,/R 2
/
/
--
I

10 25 50 15 100 250 500 150 1000 2500 5000

J r • ..,...ru,.(k) I
~ B..

Fig. 12- 16 Experimental values of t ,/R for hydrogen as a

function of temperature plotted on a logarithmic scale.

Many important featu res of the general theory have been ignored in the
( relatively) simple treatment of the problem given here. Some of these are: (a) the
difference between the behavior of molecules such as H, , whose atoms are alike,
and those, such as NO, composed of unlike atoms; (b) the degeneracy of the rota·
tional energy levels as a result of space quantization; (c) the energy associated with
electronic excitation at high temperatures; (d) the coupling between rotational
and vibrational states; and (e) the fact that the vibrations are not precisely simple
harmonic. However, the exact theory is apparently so firmly established that
specific heat capacities of gases can be computed theoretically, from optical
measurements, more accu rately than they can be measured experimentally by the
technique of calorimetry.
 380 APPLICATIONS OF STATISTICS TO GASES

PROBLEMS

12-1 In Section 12-1 the properties of a monatomic ideal gas were calculaied using the
classical d istribution funcl ion. (a) Derive the equatio n o f s1a1e and specific heat capacity
of a n ideal gas using, instead, 1he M-B dislribulion function . (b) Show tha t lhe M-B
distribulion function leads to an expression for the enlropy of an ideal gas which is not
extensive.
Il-l In a two-dimensional gas the molecules can move freely on a plane, but are con-
fined within an area A . (a) Show that the partilion funcl ion for a two-dimensional
monatomic gas of N particles is given by

A2wmkT
z--,.-.
(b) Find the equal ion of state of the gas from ils Helmholiz function.
11-3 Use the partition function of the previous problem to derive the heat capacity
and enlropy of a two-dimensional mona1omic gas.
12-4 In Fig. 12- 3, let Vz - Vw - v. - v,.., and Jet Ou.. - flu~ - Av1 ""'" O.Olu,... If N •
Avogadro's number, 6.02 x 1010 molecules, com pule the average number of pa!iicles
in each of the following elemenls of velocily space: (a) I he slice of lhickness <l.v,, (b) 1he
rectangular parallelepiped common to two slices, (c) ihe volume element Av, Au, Av.,
(d) I he sphel-ical shell of radius V3 v., a nd thickness O.OiuM.
12-5 (a) What is the "dislance" "• in Fig. 12-3, of a slice at r ight angles to ihe u,-uis, if
the slice contains one-half of the number of particles as a parallel slice of the same I hick-
ness a1 lhe origin. Express your answer in terms of •M· (b) AI what radial "diSiance" v
from the origin in velocity space is the densily p, one-half as great as I he origin.
12-6 Find the fracliona l num ber of molecules of a gas having (a) velocilies with X·
components between u., and I.Oiv,., (b) speeds between u,. an d I.Oiv.,, (c) velocities
with x-, y-, and •-components between u,. and I.Oiv.,.
11-7 Show thai v., - V2kTfm.
11-8 (a) Compute to three significant figures the rms, average, and most probable
speeds of an oxygen molecule at 300 K. (b) Compu1e 1he mosl probable speed of an
oxygen molecule a t the following temperaJUres : 100 K, 1000 K, 10,000 K.
12- 9 Show that (UI) - <0' > o. This difference plays an imporlant pau in theory of
fluctuations, and is the mean square deviation of the veloci ty from the average velocity.
12- 10 Show thai I he average reciprocal speed (1 /v) is given by 2/V; v,,. - V2mfwkT.
12- 11 (a) Express Eq . (12-18) in terms of I he kinetic energy<( - mv'/2) of the molecules.
(b) Find the most probable and average energy of molecules having a dimibution of
speeds given by Eq. (12- 18) and compare the resuiiS to mc:.0/2 and mi:2/2, respeciively.
12- 12 Show that the number of molecules wilh positive x-components of velocity less
N
l erf (x), where x = •I•.. and err (x) is the error
than some a rbi1rary value u is Ar, ... -
 PROS LEMS 381

function defined as

erf (x) - 2 f.", - •' dx.

v;
1

(b) Show that the number of moleculos with positive x-component5 of velocity larger
1han the value v is .A"_,. - ~[I - erf (x)]. Compute the fraction of moleculos with
x-components of velocity between (c) 0 and v,., (d) v,. and oo, (e) 0 and oo, (f) - • .. and
+v,.. The value of erf ( I} - 0.8427. (g) Illustrate your answers graphically in terms of
the velocity distribution function.
12- 13 Show that the number of molecules with spuds lw than some arbit rarv value v
is given by

where x and err (x) are defined in the p revious problem. (b) Show that the number o f
molecules with speeds greater than the arbitrary value is given by

Compute the fraction of molecules with speeds between (c) 0 and v.,., (d) v,. and oo, and
(e) 0 and oo. (f) Illustrate your answers graphically in terms of the speed distribution
function.
12-14 Show that •..... for particles leaving a small hole in a furnace is given by ,;4k1]m.
12-IS Show that the number of molecules colliding with a surface of unit area per unit
time, with components of velocity at right anglos to the surface greater than some a rbi trary
value v = xv.., is [nv,. exp ( - x'}J/2v;.
12- 16 The oven in Fig. 12- 10 contains bismuth at a temperature of 830 K, the drum
is IOcm in diameter and rotates at6000rpm. Find the distance between the point$ of
impact of the molecules Bi and Bi, on the glass plate, G. Assume that all the molecules
of each species escape the oven with the rms speed appropriate to tha t species.
12- 17 A sph<rical bulb 10 em in radius is maintained at a temperature of 27°C, except
for one square centimeter , which is kept a t a very low tempera ture. The bulb contains
water vapor originally at a pressure of 10 Torr. Assuming that every water molecule
striking the cold area condenses and sticks to the surrace, how long a time is required for
the prwure to decrease to Ia-' Torr ?
12-18 A spherical bulb IOcm in radius is pumped continuously to a high vacuum. In
the bulb is a small vessel, closed except for a circular hole 0.2 mm in diameter located at
the center of the bulb. The vessel contains mercury at I00°C, at which temperature its
vapor pressure is 0.28 Torr. (a) Compute the average speed r of the molecules of mercu ry
vapor in the small vessel. (b) Compute the rate of effiu• of mercury through the hole, in
milligrams hr-1• (c) How long a time is required for I microgram of mercury to be
 382 APPLICATIONS OF STATISTICS TO GASES

To pump

Figure Jl.-17

deposited on a square centimeter of the inner surface of the bulb, in a direction making
an angle of 4S 0 with the normal to the hole? (Sec Fig. 12-17.)
11-19 In a molecular beam experiment, the souroo is a tube containing hydrogen at a
pressure P, - O.I STorr and at a temperature T - 400 K. In the tube wall is a slit
30 mm >< 0.025 mm, opening Into a highly evacuated region. Opposite the source slit
and I meter away from it is a second detector slit parallel to the first and of the same size.
This slit is in the wall of a small enclosure in which thcpressu rc P 0 can be measured. When
the steady state has been reached: (a) What is tho discharge rate of the source slit in micro-
grams s"""'? (b) What Is tho rate of arrival of hydrogen at the detector slit, in micro-
grams s-1, and in molec:ulos s- •1 (c) How many molec:ulos that will eventually
reach the detector slit arc in tho space between source and detector a t a ny instant? (d)
Wha t is the equilibrium p ressure P0 in the detector chamber?
12-20 Tho distanocs OS and SD in the appara tus of Estcrmann, Simpson, and Stem in
F ig. 12- 11, are each I meter. Calculate the distance of the detector below the central
position D, for cesium atoms having a speed equal to the rms speed in a beam emerging
from an oven a t a temperature of 460 K. Calculate also the "angle of elevation" of the
trajectory. The atomic weight of ocsium is 133.
12-21 The neutron flux across an area at the center of the Brookhaven reactor is about
4 >< 10" neutrons m-ts-1• Assume that the neutrons have a Maxweii-Boltunann
velocity dist ribution corresponding to a temperature of 300 K ("thermal" neutrons).
(a) Find tho number of neutrons per cubic meter. (b) Find the "partial pressure" of the
neutron gas.
12-11 Derive Eq. (12-27) from Eq. (7-31) assuming £ 0 = rg, Y1 -nand X 1 - L.
12-13 (a) Obtain the expressions for Z , and z. given in Eqs. (12-31) and (12-32). (b)
Obtain the expressions for nand r given in Eqs. (12- 34) a nd (12- 3S).
12- 14 For the gas in a cylinder in a gravitational field, discussed in Section 12-4, show
that as g-+ 0, the number of molecules per unit volume approaches the constant value
N/ V, and hence is the same at all elevations. In other words, in Ihe absence of a gravita·
tiona! field tho molecules of a gas a re distributed uniformly throughout the volume of a
container.
12-lS Show that the net downward force exerted on tho container by Ihe gas, in Section
12-4, equals tho weight of the gas in the container.
 PROBUMS 313

12-26 If the height of the atmosphere is very large. show that (a) n- 0, (b) r•
s
NkT/g. (c) E = 2NkT, (d) dS = Nk[(S/2)(dT/T)- (dg/g)J, and (e) that states at
constant entropy arc related by 7"12/g - constant.
ll-17 (a) Calculate the fraction of hydrogen atoms which can be thermally ionized at
room temperature. (b) At what temperature will r 1 of the atoms be ionized 7
11- 28 When a gas is whi rled in a centrifuge, its molecules can be considered to be acted
on by a radially outward centrifugal force of magnitude moh. Show that the density
of the gas as a func tion of r varies as exp (m•h'/2kT). ·
11-29 Find the mean gravitat ional potential energy per molecule in an infinitely h igh
~ isothermal atmosphere.
U-30 (a) Usc the principle of equipartition of energy to flnd the total energy, the energy
per particle, and the heat capacity of a system of N distinguishable harmonic oscillators in
equilibrium with a bath at a temperature T. The kinetic energy of each oscillator is
mM + v: + v:l/2 and the potential energy is X(x' + y1 + z'J/2 wherex,y, and zare the
displacements from an equilibrium position. ( b) Show that the expansivity of this system
is zero because X • j - l • 0.
ll-31 A molecule consists of four a toms at the corners of a tetrahedron. (a) What is the
number of translational, rotational, and vibrational degrees of freedom for this molecule?
(b) On the basis of the equipartition principle, what arc the values of c. and y for a gas
composed of these molecules 1
l l-31 Using Eq. (1 1-62) derive (a) Eq. (12-48) and (b) Eq. (12-49). (c) Show that when
T » 9, Cv approaches Nk; and when T « 9, Cv approaches zero as e-41'1'.
11- 33 Calculate the average fractional number of oscillators in thejth energy level N1/N
for the four lowest energy levels when (a) T • 9/2 and (b) T - 29.
11-34 Make sketches of the average fractional number of oscillators in (a) the ground
state, and (b) the flrst excited state, and (c) in the second excited state as a function of
Tf9.
12~35 Making usc of Eq. (1 1-66), show that the entropy of an assembly of quantized
linear oscillators is

S • Nk {exp (::~ _ 1 - In (I - cxp ( - 9/TJJ}.

where 9 • hrjk. (b) Show that Sapproaches zero as Tapproaches zero. (c) Why should
Eq. (11- 66) be used rather than Eq. (1 1-63)?
1
11- 36 Consider 1000 diatomic molecules at a temperature 9, 1,)2. (a) Ftnd tHe number
in each of the three lowest vibrational states. (b) Find the vibrational energy of the
system.
! .
 13
. I
Applications of quantum statistics to
other systems
13- 1 THE EiNSTEIN THEORY Of THE SPECIFIC HEAT CAPACITY Of A SOLI D

13-2 THE DEBYE THEORY Of THE SPECIFIC HEAT CAPACITY OF A SOLIO

13-3 BLACKBODY RAOIATION

13-4 PARAMAGNETISM

13-5 NEGATIVE TEMPERATURES

13-6 THE ELECTRON GAS

 385 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13- t

13-1 THE EINSTEIN THEORY OF THE SPECIFIC HEAT

CAPACITY OF A SOLID
In Section 9-8 and Fig. 3-10 it was shown that the specific heat capacity of many
solids, at constant volume, approaches the Dulong-Petit value of 3R at high tem-
peratures, but decreases to zero at very low temperatures. The first satisfactory
explanation of this behavior was given by Einstein, who proposed that the atoms
of a solid be considered in the fi rst approximation as an assembly of quantized
oscillators all vibrating with the same frequency 7. The principles of quantum
mechanics had not been completely developed at the time this suggestion was made,
and Einstein's original article assumed that the energy of a n oscillator was given by
~, -= n1h•.
The additional factor 1/2, which we introduced in Eq. ( 12-43), does not
affect the method and we shall make usc of the expressions already derived in
Section 12-6. We must make one change, however. T he atoms of a solid are free
to move in thru dimensions, not just one, so that an assembly of N atoms is
equivalent to 3N linear oscillators. Then from Eq. ( 12-48), the internal energy U
of a solid consisting of N atoms is

U .. 3Nkllr: [ I
exp (lla/T) - I
+2l]. (13-1)

where the Einstein ttmperature 1111 is defined as

/lr. E!!! . (13-2)

k
The mean energy of an atom is

l =- !!_ - 3kll 11 [ I + !l,

N exp (lle/T) - I 2J
and the specific heat capacity at constant volume is

c = 3R(~)" exp (lla/T) (13-3)

• T [exp (/lefT) - 1]1 •
Figure 13-1 shows graphs of the dimensionless ratios </kllr. and cJ R, plotted
as functions o f Tj/1.,_. The ordinate of the latter curve, at any temperature, is pro-
portional to the slope of the fo rmer. The general shape of the graph of c. is in
agreement with the experimental curve shown in Fig. 3-10. The value of 118 (and
hence of 7) for a particular substance is chosen so as to get the best fit between
theoretical and experimental curves. However, it is not possible to find a value
of liE that gives good agreement at both low and high temperatures.
»
When T 11.:. II,.JT is small and c. approaches the Dulong-Petit value,
c. .. JR.
 13- 2 THE DEBYE THEORY OF THE SPECIFIC HEAT CAPACITY OF A SOLID 387

Fig. 13-1 Internal energy and specific

heat capacity of a harmonic oscillator.

When T «OE, the exp onential term is large, we can neglect the in the de-
nominator, and

(See Prob. 12-32.)

When T approaches zero the exponential term goes to zero more rapidly than
1/T' goes to infinity, and c. approaches zero in agreement with experiment and
the third law. H owever, because of the rapid decrease of the exponential term, the
theoretical values of c., at very low te mperatures, decrease much more rapidly
tha n the experimental values. Thus the Einstein theory, while it seems to indicate
the correct approach to the problem, is evidently not the whole story.

13- 2 THE DEBYE THEORY OF THE SPECIFIC HEAT

CAPACITY OF A SOLID
The simple Einstein theory assumes that all atoms of a solid oscillate at the same
frequency. N ernst and Lindemann• found empirically that the agreement between
theory a nd experiment could be improved by assu ming two groups of atoms, one
oscillating at a frequency • and the other at a frequency 2v. This idea was extended
by Born,t von Karman,! and Debye, who considered the a toms, not as isolated
oscillators all vibrating a t the same frequency, but as a system of coupled oscillators
having a continuous spectrum of natural freq uencies.

• Frederick A. Lindemann, First Viscount Cherwell, British physicist (1886-19p7).

t Max Born, German physicist (1882- 1970).
t Theodor von Karman, Hungarian engineer (1881- 1963).
 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13-2

As a simple example of coupled oscillators, suppose we have two identical

particles connec te~ by identical springs, as in Fig. 13-2. If both particles are given
equal initial velocities in the same directio n, as indicated by the upper arrows, the
particles will oscillate in phase with a certai n frequency v1 • If the initial velocities
are equal and o pposite, as indicated by the lower arrows, the particles will oscillate
o ut of phase but with a different frequency v,. If the initial velocities have arbitrary
val ues, the resultant motion is a superposition of two oscillations of frequencies
v1 and ••· The system is said to have two naturalfr~qu~nci~s.
I

Fla. 13-2 Coupled oscillators.

Now suppose that the number of particles (and springs) is increased. It is no

great tas k to calculate the natural frequencies when the num ber is small, but as
the number is increased there a re too many simullaneous equations to be solved .
It turns out, however, tha t if there are N particles m the chain, the system will have
N natural frequencies, whatever the value of N
N ow extend these ideas to three dimensions. A simple model of a c rystal
consists of a three-dimensional array of particles connected by s prings, a nd such
an array has JN natural frequencies. Because of the impossibility of calculating
these frequencies when N is as large as the number of molecules in a macroscopic
crystal, Debye assumed that the na tu ral frequencies of the atoms of a crystal would
be the same as the frequencies of the possible stationary waves in a crystal if the
crystal were a continuous elastic solid. This is a standard problem in the theory of
elasticity, and we shall outline its solution without giving details. The procedure
is closely analogo us to that described in Section 11-2, except that we are now
dealing with real elastic waves and not with the mathematical waves of wave
mechanics.
As explained in Section 11-2, an elastic string of length L fixed at both ends,
can oscillate in a steady state in a ny mode for which the wavelength A is given by

A-!!:
n '
where n = I, 2, 3, •.. , etc.
The fundamental equation o~ any sort of wave motion states that the speed of
propagation c equals the product of the frequency v and the wavelength l:

c- •.t
 13-2 THE OEBYE THEO RY OF THE SPECIFIC HEAT CAPACITY OF A SOLID 389

It follows that for a ny frequency v, the number n is

2L
n=-l'
c
a nd
n• = 4L' ~·.
c•
The theory of elasticity leads to the result tha t the natural frequencies of
statio na ry waves in a n elastic solid in the form of a c ube of side length L are given
by the same equation except tha t the possible values of n' nre

n1 . . n: + n: + n!.
where n., n,, and n, are positive integers that can have the values I, 2 , 3, . J.,
etc.
To lind the number of waves in any frequency interval, o r the frtqutncy
sputrum, we proceed in t he same way as in Section 12-1 and Fig. 12-1. Let the
numbers n,, n, , and n, be laid off on three mutually perpendicular axes. Each
triad of values determines a point in n·space, wi th corresponding values of n a nd
of~. Let '§ represent the total number of possible frequencies, up to and including
that corresponding to some given n. This is equal to the number of points withi n
a n octant o f a sphere of radius n, the volume of which is (11/6) n', and since n -
(2Lfc)v,

But L' is the volume V of the cube, and it can be shown that regard less of the
shape of the solid we can replace L' with V. Then

'§= ~!:: ~·. (13-4)

3 c'
However, three types of elastic waves can propagate in an elastic solid : a
longitudinal or compressional wave (a sound wave) traveling wi th speed c1 , and
two transverse or shear wa ves polarized in mutually perpendicular directions and
traveling with a differe nt speed c1• T he total number of possible statio nary waves
having frequencies up to and including some frequency~ is the refore

'§ - ~
3
v(.!. + c!.!.)~•.
c~
(13-S)

According to the Debye theory Eq. ( 13-S) can a lso be interpreted as de·
scribing the number of linear oscillators havi ng frequencies up to and incl udin g the
frequency ~. Th us, to be consistent with the notation of Section 12-2, ~ in Eq.
(13-5) should be replaced by .A'" and

.A'" = ~
3
v(.c~!. + 1)•'·
c~
 3tO APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13-2

If there were no upper limit to the frequency, the total number of oscillators
would be infinite. But a crystal containing N atoms constitutes an assembly of
3N linear oscillators. Hence we assume that the frequency spectrum cuts off at a
maximum frequency '• such that the total number of linear oscillators equals 3N.
Then setting 5 - 3N and • = '••

(13-7)

The wave speeds c1 and c, can be calculated from a knowledge of the elastic
properties of a given material and hence '• can be calculated from this equation.
In a material like lead, which is easily deformed, the wave speeds are relatively
small, while in a rigid material like diamond, the speeds are relatively large. Hence
the value of,,. for lead is much smaller than it is for diamond.
That there should be a maximum frequency of the stationary waves that can
exist in a real solid can be seen as follows. For a single set of waves of speed c,
the maximum frequency "• corresponds to a minimum wavelength .<..,. = cfv.,
and Eq. (13-7) can be written

AmiA-9 (4")'11("N'
Y)'" (13-8)

But (V/N) is the average volume per atom and the cube root of this, (V/N)•h, is
of tbe order of the average interatomic spacing. Hence the structure of a real
crystal (which is not a continuous medium) sets a limit to the minimum wave-
length which is of the order of the interatomic spacing, as wo uld be expected since
shorter wavelengths do not lead to new modes of atomic motion. It follows from
Eqs. (13-6) and (13-7) that

The number of linear oscillators having frequencies between v and ,. + A• is

then
A%, - ~ ,• a •.
9
...
a nd the number per unit range of frequency is
(13-9)

A%, 9N 1
(13-10)
~-,.~"·
Figure 1~-3 is a graph of A%,/Av, plotted as a function of • · The actual
number A.#l, of osciUators of frequency between • and v + A• is represented by
the area of the shaded vertical strip, since the height of the strip is a%,ft.v and
its width is A•. This is in contrast to the Einstein model, in which all oscillators
have the same frequency. The total area under the curve corresponds to the total
number of linear oscillators, 3N.
 13-2 THE DEBYE THEORY OF THE SPECIFIC HEAT CAPACITY OF A SOUD 3t1

Fig. 13-3 The Debye frequency

spec1rum.

The oscillators of any frequency~ constitute a subassembly of linear oscillators

all having the sam~ frequency , as in the Einstein model. .Then from Eq. (12-48)
the internal energy l!.U, of the subassembly, replacing 3N with l!..;V., is

~
0
l!.U 9N h• 1:!.• (13-11)
' •!. exp (hv/k T) - 1 ·

We omit the constant zero·point energy since this has no effect on the heat capacity.
The point of view thus far, in this section and in the preceding one, has been
to consider the atoms of a crystal as distinguishable particles obeying the M-B
statistics. An alternative approach is to consider the elastic waves themselves as
the "panicles" of an assembly. Each wave can also be considered as a particle
called a phonon, and the assembly described as a phonon gas. Since the waves or
phonons are indistinguishable, and there is no restriction on the number permitted
per energy state, the assembly obeys the Bose-Einstein statistics.
We must, however, make one modification in the expression previously
derived for the distribution function in this statistics. This is because the number
N of waves, or phonons, in contrast to the number of atoms of a gas in a container
of specified volume, cannot be considered one of the independent variables speci-
fying the state of the assembly. If the assembly is a gas, we can arbitrarily fix both
the volume V and the temperature T of a container, and still can introduce any
arbitrary number N of molecules of gas into the container. But when the volume
and temperature of a crystal are specified, the crystal itself, so to speak, deter-
mines the number of different waves, or phonons, that are equivalent to the oscilla-
tions of its molecules. Thus the crystal cannot be considered an open system for
which N is an independent variable and the term JJ dN does not appear in Eq.
(11-22). This is equivalent to setting JJ = 0 and hence exp (JJ/kT) = I. The
 392 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13-2

number of "particles" in a macrolevel between • and • + t:u is therefore

ll..A' = _ _ll.-:---
~-­ ( 13- 12}
exp (•/kT) - 1
According to the principles of quantum mechanics, the energy of a wave (or
phonon) of frequency • is

where h is Planck's constant. Unlike a linear oscillator of frequency •, which can

have any one of the energies (n 1 + i )h•, where n1 - 0, I, 2, ... , etc., a wave of
frequency • can have only the energy h•. Thus if a large amount of energy is
associated with a given frequency, this simply means that a large number of waves,
or phonons, all of the same energy, are present in an assembly.
An energy interval between • and • + ll.< therefore corresponds to a frequency
interval between • and • + ll.•. Thus the number of phonons with frequencies
between • and • + cb i s
ll..A' ll. (9, (13- 13)
'~ exp(h/kT}- 1 '
where ll.~, is the number of states having frequencies between • + • + d•.
The energy ll.U, of the waves in this frequency interval is

ll.U h ll..A' hvll.~,

' - • ' - exp (h•/kT} - 1 '
and comparison with Eq. (1 3- 11) shows that

ll.~.~9~··ll.••
·- (13-14)

which is the same as the expression for the number of distinguishable oscillators
in this frequency interval. That is, the degeneracy ll.~, of a macrolevel is equal to
the number of distinguishable oscillators in the same interval. Equation (13-13)
can therefore be written
ll..A', - 9 N v' ll.v . (13 15)
·~ exp (hvfkT) - 1 -
There appears at first sight to be a discrepancy between the expression for
ll..A', in the preceding equation and that in Eq. ( 13- 9). However, the symbol
ll..A', docs not represent t he same thing in the two equations. In Eq. ~13- 1 5),
ll..A', is the number of indistinguishable waves (or phonons) having frequencies
between • and • + ll.v, in a system obeying the B-E statistics. In Eq. ( 13-9),
ll..A', is the number of distinguishable oscillators having frequencies in the same
range, in a system obeying M-B statistics.
 13-2 THE DEBYE THEORY OF TH E SPECIFIC HEAT CAPACITY OF A SOLID 393

T he to tal energy U of t he assembly is now obtained by summing the expression

for flU. over all values of • from zero to "•• and after replacing the s um with an
integral, we have

u =9Ni•"'
-
h.. d•. ( 13-16)
•!. o exp (h/ kT) - I
The De bye ttmperoturt 80 is defined as

8 - h•.. ( 13-17)
o=T•
and 80 is proportional to che cut-off frequency • ... Some values are given in Table
13-1.

Table 13-1 Debye temperatures of

some materials

Substance BD(K)

Lead 88
Thallium 96
Mercury 97
Iodine 106
Cadmium 168
Sodium 172
Potassium bromide 177
Silver 21S
Calcium 226
Sylvine (KCI) 230
Zinc 235
Rocksah (NaO) 281
Copper 315
Aluminum 398
Iron 453
Fluorspar (Ca F,) 474
Iron pyrites (FeS,) 645
Diamond 1860

For convenience, we inlroduce the dimensionless quantities

h> h•., 80
x= - X~ - ~ -

kT' • kT T.
Then

U T )'i'• x' dx
= 9NkT- ( 13-18)
(8 o exp (x)- 1
0
 314 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13-2

This corresponds to Eq. (13-1) for the energy U according to the Einstein theory.
Consider lint the high temperature limit, at which x • h>/kT is small. Then
[exp (x) - IJ ~ x and the integral becomes

Then at high temperatures,

f.·-
o
x1 dx ::a
x•
3
o•
-.! - _!!_1 •
3T

U ~ 3NkT, c.- 3R.

in agreement with the Einstein theory and the Dulong-Petit law.
At intermediate and low temperatures, the value of the integral can be expressed
only as an infinite series. To a good approximation, the upper limit of the integral
when Tis very small can be taken as infinity instead of x,., since the integrand is
small for values of x greater than x.,. The definite integrallhen equals "'/IS, and
hence at low. temperatures,
3
S
(
U=-.,'NkT- · T)'
00 '
and by differentiation,
1211. ( T_\1
c. - -s- R B;r (13-19)

Equation (13-19) is known as the Debye T' law. According to this law, the
heat capacity near absolute zero decreases with the cube of the temperalure,
inslead of exponentially as in the Einslein theory. The decrease is therefore less
rapid and !he agreement with experiment is much beller. Although !he Debyc
theory is based on an analysis of elastic waves in a homogeneous, isotropic, con-
tinuous medium, experimental values of the specific heat capacity of many
cryslalline solids are in good agreement with the Debye theory at temperatures
below 00 /50, or when T/00 < 0.02. As the !emperature is increased, the specific
beat capacity increases somewhat faster than the lheory would predict. There is
recent experimental evidence that amorphous malerials do not appear to follow
!he Debye 71' law even a11empera1ures below 00 /100, or when T/00 < 0.01.
The heal capacity at any temperalure can be calculated by evalualing the inle·
gral in Eq. (13-18), which gives !he internal energy as a funclion of T, and differen-
liating !he result wilh respect toT. As in the Einstein theory, lhe resull is a function
of Tf00 only, and hence a single graph represents the tem perature varia lion of c.
for all substances. The curve in Fig. 13-4 (what can be seen of it) is a graph of
c.fR , plotted as a function of T/00 , and the points are experimental values for a
variely of materials.
It will be seen from Fig. 13-4 that roughly speaking, when T/00 is grealer !ha n
I, or when the actual temperature exceeds the Debye temperature, the syslem
behaves "classically" and c. is nearly equal to !he "classical"' or "non-quantum"
value JR. When the actual temperalure is less than the Debye tempera!Ure,
 13-3 BLACKBODY RADIATION 396

xl'b e.~ 88 K

.I
. • Aa
"KCJ
o Zn
0 N•CJ
21S
230
2JS
21!1
6Cu JIS
,t +AI 398
:- o
.c
C.F1 474
1860
;
..I o.s 1.0 I.S lO 2.S
TJ(J.

Fig. 13-4 Specific h~at capacities of various solids as functions of T/90 .

quantum effects become significant and c. decreases to zero. Thus for lead, with a
Debye temperature of only 88 K, " room temperature" is well above the Debye
temperature, while diamond, wirh a Debye temperarure of 1860 K, is a "quantum
solid" even at room temperature.
At intermediare temperatures there is good agreement between values of the
specific heat capacity calculated by the Einstein and by the Debye theories. This
agreement might be expected, since the Dulong-Petit theory is a first approximation
that works at high temperatures. The Einstein theory is a second approximation
which works for high and intermediate temperatures. The Debye theory is a third
approximation that works at low temperatures when other effects do not dominate.

13- 3 BLACKBODY RADIATION

The thermodynamics of blackbody radiation was discussed in Section 8- 7 and

we now consider the statistical aspects of the problem. The radiant energy in an
evacuated enclosure whose walls are at a temperature T is a mixture of elecrro-
magnetic waves of all possible frequencies from zero to infinity, and it was the
search for a theorerical explanation of the observed energy distribution among
these waves that led Planck to the postulates of quantum theory.
To apply the methods of sratistics to a batch of radiant energy, we consider
the waves themselves as rhe "particles" of an assembly. Each wave can be con-
sidered a particle called a photon and the assembly can be described as a photon
gas. Because the photons are indistinguishable and there is no restriction on the
number per energy state, the assembly obeys the Bose-Einstein statistics.
 396 APP~ICATIONS OF QUANTUM STATISTICS TO OTH ER SYSTEMS 13-3

The problem is very similar to that of a phonon gas discussed in the preceding
section. The number of photons in the enclosure cannot be considered an inde-
pendent variable and the B-E distribution function reduces to the simpler form,
ll.IV t:. r§,
• = exp (h~fkT) - 1
There is, however, a difference in the expression for the degeneracy t:.r§,.
As we showed in the preceding section, the degeneracy of a macrolevel, in an
assembly of waves (or photons) is equal to the possible number t:.t'§, of stationary
waves in the frequency interval from • to v + t:.v. Let us return to Eq. (13-5),

<'§~~!.•
J ,~ ,
where<'§ is the number of stationary waves with frequencies up to and including~.
Electromagnetic waves are purely transverse and there can be toro sets of waves,
polarized in mutually perpendicular planes and both traveling with the speed of
light c. Also, since empty space has no structure, there is no upper limit to the
maximum possible frequency. Then interpreting (1 as the total number of possible
energy states of all frequencies up to and including '• we have

r§-~K .•.
3 c•
The degeneracy t:.r§, is therefore
t:.r§ 81rV 't:.
·-7" Y,

and the number of waves (or photons) having frequencies between v and • + t:.v
is
t:.% , • 81rV •' t:.v. (13- 20)
c' exp (h•/kT) - 1
The energy of each wave is hv, and after di viding by the volume V, we have
for the energy per unit volume, in the frequency range from • to • + t:.~. or the
spectral energy dmsity t:.u.,
t:.u, = 87rh •'
c• exp (h•/kT) - 1
t:... (13-21)

This equation has the same form as the experimental law (Planck's law) given in
Section 8-7, and we now see that the experimental constants r, and c, in Eq.
(8-50) are related to the fundamental constants h, c, and k, by the equations
h
c, ... k. (13- 22)
 13-3 BLACKBODY RADIATION 397

When numerical values of h, c, and k are inserted in these equations, the cal·
culated values of c1 and c, agree exactly with their experimental values, within the
limits of experimental error.
At a given temperature T, and at high frequencies for which hv)) k T, the
exponential term is large; we can neglect the I; and

t.u. ~ S~h v' exp ( -hvfkT) C.~. (13-23)

c
An equation of this form had t>et.n derived by Wien• before the advent of quantum
theory and it is known as Wien's !all'. It is in good agreement with experiment at
high frequencies but in very poor agreement at low frequencies.
«
However, at low frequencies for which hv kT, [exp (hrfk T) - I] is very
nearly equal to hr/k T and
8.,.kT ,
Au.. ~T~ Av. ( 13-24)

This equation had been derived by Rayleight and Jeans,! also before the quantum
theory, and had been found to agree with experiment at low, but not at high, fre·
quencies. That it cannot be correct in general can be seen by noting that as the
frequency becomes very high, the predicted energy density approaches infinity.
(This result is sometimes referred to as the "ultraviolet catastrophe.")
It is interesting to note that Planck's first approach to the problem was purely
empirical. He looked for an equation having a mathematical form such that it
would reduce to the Wien equation when hrfk T was large, and to the R,ayleigh·
Jeans equation when hrfkTwas small. He found that Eq. ( 13-2 1) had this p~operty,
and his search for a theoretical explanation of the equation led to the development
of quantum theory.
Figure 13- 5 shows graphs of the dimensionless quantity ~:·(8 ;:~~), plotted
as a fu nction of the dimensionless quantity hvfk T. The solid curve is a graph of
Planck's Jaw, and the dotted curves are, respectively, graphs of the Rayleigh·
Jeans Jaw, applicable when /~r « kT, and of Wien's Jaw, applicable when
hv)) kT.
The total energy density u., including nil frequencies, can now be found by
summing t.u. over all values ofr from zero to infinity, since there is no limit to the
maximum value of v. Replacing the sum with an integral, we have

Uy
g.,."L"- ••
=---.., dJI;
c o exp (lr •fkT) - I
• Wilhelm Wien, German physicist (1864-1928).
t John W. Strull, Lord Rayleigh, English physicist (1842- 1919).
t Sir James H. Jeans, English mathematician (1877- 1946).
 398 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS t3-3

Planck

Fig. 13- 5 Graphs of Planck's Jaw, Wicn's Jaw, and lhc

Rayleigh-Jeans Jaw.

or, if we define a dimensionless variable x = ilv/kT,

u - 87Tk'
' - c3h3
r•J...expx•(x)dx-
o I

The value of the definite integral is 7T'/I S, so

g,'k'
u, = 15c'h' T' = uT', (IJ.-25)
where
g,'k'
u---.
15c h 3 3
(JJ- 26)

Equation (IJ.-25) is the same as Stifan's law (Eq. (8-54)); and when the values of
k, c, and hare inserted in Eq. (13- 26), the calculated and experimental values of"
agree exactly, within the limits of experimental error.
Thus quantum theory and the methods of statistics provide a theoretical basis
for the form of Planck's law, and relate the experimental constants r 1, r,, and "
to the fundamental constants II, c, and k. Expressions for the internal energy, the
entropy, and the Helmholtz and Gibbs functions of blackbody radiant energy were
derived by the principles of thermodynamics in Section 8-7 and need not be
repeated here. It will be recalled that the Gibbs function G = 0, which might also
have been taken as a justification for setting}' • 0 in the B-E distribution function.
 13-4 PARAMAGNETISM 398

13-4 PARAMAGNETISM
We now conside r the statistics of a pa ramagne tic crystal. The properties of such
crystals a re chiefly of interest in the region of extremely low temperatu res, of the
o rder of a few kelvins or less. A number of simplifying assumptions will be made,
but the procedure is the same as in more complicated cases.
A typical paramagnetic crystal is chromium potassium sulfate, Cr,(SO, ).-
K,S0,24H,O. Its paramagnetic properties a re due solely to the chromium atoms,
which exist in the crystal as ions, Cr+++. Every electron in a n a tom has not only an
electric charge but also a magnetic moment /Ln of I Bohr* magneton, equal (in
MKS units) to 9.27 x 10- u A m', as if the electron were a tiny sphere of electric
charge spinning about an axis. In most atoms, the resultant magnetic moment of
the electrons is zero, but the chromium ion Cr+++ has th ree uncompensated electron
spins and a magnetic moment of 31'n·
For every chromium ion there a re 2 sulfur atoms, I potassium atom, 20
oxygen a toms, and 24 hydrogen atoms, making a total of 47 other particles which
a re nonmagnetic. The magnetic ions are therefore so widely separated that there
is only a small magnetic interaction between them.
It was shown in Section 8- 8 that the thermodynamic properties of a para-
magnetic crystal could be calculated from a knowledge of the quantity F* ~
E - TS. Using the methods of sta tistics, the expression for F* can be derived in
terms of the temperature Tand the parameters tha t determine the energy levels of
the atoms in the crystal. Because the atoms can be labeled according to the
p ositions they occupy in the crystal lattice, the system obeys M-B statistics, and as
usual the first step is 10 calculate the partition function Z, defined as

Z =Ill. '#1 exp-!!....

1 kT
Because of their oscillatory motion, the molecules have the same set of vibra-
tional energy levels as those of any solid, and the total vibrational energy con-
stitutes the internal energy U,.,.. I n addition, the small interaction between the
magnetic ions, and their interactions with the electric field set up by the remainde r
of the la ttice, gives rise to an additional internal energy (of the ions only) which we
write as u,.,. Finally, if there is a magnetic field in the crystal, set up by some
external source, the ions have a magnetic potential energy which, like the gravita-
tional potential energy of particles in a gravitational field, is a joint property of
the ions and the source of the field and cannot be considered an internal energy.
The total magne tic potential energy is £,,.
The vi brational energy levels, the levels associated with internal magnetic and
electrical interactions, and the potential energy levels a re all independent. The
partition function Z, as in the case of a gas in a gravitational field, can be expressed

• Niels H. D. Bohr, Danish physicist (1885-1962).

 400 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13-4

as the product of independent partition functi ons which we write as Z,.,•• z,.,, and
Z". Thus
z = z•••z,.,z....
The magnetic ions constitute a subassembly, characterized by the partition
functions z,.,
and Z"' only, and they can be considered independently of the
remainder of the lattice, which can be thought of simply as a container of the sub-
assembly. A lthough the energy u,., and the partition function Z1., play important
roles in the complete theory, we shall neglect them and consider that the total
energy of the subassembly is its potential energy £• only. Thus we consider only
the partition function Z"'.
As shown in Appendix E, the potential energy of an ion in a magnetic field of
intensity Jf' is -p..Jf' cos 0, where p. is the magnetic moment of the ion and 0 the
angle between its (vector) magnetic moment and the direction of the field. For
simplicity, we consider only a subassembly of ions having a magnetic moment of
I Bohr magneton Jl.n· The principies of quantum mechanics restrict the possible
values of 0, for such an ion, to either zero or 180°, so that the magnetic mome nt is
either parallel or antiparallel to the field. (Other angles are permitted if the
magnetic moment is greater than p. 0 ). The corresponding values of cos 0 a re then
+I and -I, and the possible energy levels are -p. 0 Jf' and +p. 0 .Jf'. The energy
levels are nondegenerate ; there is only one state in each level, but there is no
restriction on the num ber of ions per state. The partition function Z"' therefore
reduces also to the sum of two terms:

ZA' = exp (p.n.Jf') + exp { -p.n.Jf') = 2 cosh Jl.nJt', (13-27)

kT kT kT
since by definition the hyperbol ic cosine is given by
cosh x = ![exp (x) + exp ( -~)).
Let N 1 and N 1 represent respectively the number of ions whose moments are
aligned paral1el and anti parallel to the field Jt'. The corresponding energies are
•r = -p.0 .Jf' and 'I = p 8 Jf'. The average occupation numbers in the two
directions are then
N - •1 N 'I
N1 = - exp- N1 = -exp -.
z kT' Z kT
The excess of those ions in the parallel, over those in the an tiparallel alignment,
is

Rt - Nt = -N[exp (-
Z
-•t) -
kT
(
kT Z
pn-J('
N 2 sinh -
exp -•1 )] =- -
kT '
which reduces to

(13-28)
13-4 PARAMAGNETISM 401

The net magnetic moment M of the crystal is the product of the magnetic
moment p 11 of each ion and the excess number of ions aligned parallel to the field.
Then
n n f'uJf'
M = (IYI ~ tvl)f' u = Np 11 1anh kT. ( 13-29)

This is the magnetic tqualion of s1a1e of the crystal, expressing the magnetic
moment M as a fu nc tion of Jf' and T. Note that M depends only on the ratio
Jt'"fT.
The equation of slate can also be derived as follows. The function F * is

F* = -NkTJnZ = -NkTin [2cosh"';A· ( 13-30)

The magnetic moment M, which in this case corresponds to the extensive variable
X, is
M = - (oF")
o.Jr T
= Np 11 tanh f'n.11".
kT
(13-3 I)

I n strong fields and a t low temperatures, where p 0 .Jt')) kT, tanh (p 0 Jf'(kT)
approaches I and the magnetic moment approaches the value
( 13- 32)
But this is simply the sa1ura1ion magnetic mommt M..,. which would result if all
ionic magnets were parallel to the field.
At the other extreme of weak fields and high tempera tures, p 0 .Jt' « kT,
tanh (p 0 .1t'fkT) approaches p 11.Jt'(kT, and Eq. (13-3 1) becomes

M = ( N::•y;. (13-33)

But this is just the experimentally observed Curie law, slating that in weak fields
a nd at high temperatures the magnetic moment is directly proportional to (£'/T),
or
I
(13- 34)

where Cc is the Curie constant. The methods of statistics therefore not only lead
to the Curie law, but they also provide a theoretical value of the Curie constant,
namely,
Np~
Cc = - k-. ( 13-35)

Workers in the field of paramngneaism customarily use cgs units. The unit of
magnetic intensity is I oersted' [(I Oe) equal to to-• A m'.J The Bohr magneton is
Pu - 0.927 x to-•• erg Qe- 1,
• Hans C. Oersted, Danish physicist (1777-1851).
 402 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13--4

and the Bollzmann constant is

k - 1.38 x Jo- 11 erg K- 1•
Ir the number of part icles is Avogadro's number N•• equallo 6.02 x IOU cgs units,
the Curie constant as given by Eq. (13-33) is
N.S~t ·
Cc - - k - - 0.376 em' K mole- •.

T he complele theory leads 10 tho resull that for chromium ions cr+++, of magnetic
moment 3S~th the value of Cc is S times as greal, or
Cc - S x 0.376 - 1.88 em' K mole- •.
The experimentally measured value is
Cc - 1.84 em' K mole-•
in good agreement with the predictions of quantum theory.
The ratio M/ M,., is
..!!.._ = tan h l'n.lf'. (13-36)
M .., kT
Figure 13-6ps a graph of the magnttizalion cur~ of the system, in which the ra tio
M / M,.. is plotted as a function of p 0 .1f'/kT. The magnetization cur ve represents
the ba la nce struck by the system between the ordering effect o f the external fie ld
Jf', wh ich is to a lign a ll ioni1= magnets in the direction of the field, a nd the dis-
ordering effect of thermal agitation, which increases with temperature. In weak

Flg. l l-6 Magnelizalion curve of a para-

magnetic crystal.
 13-4 PARAMAGNETISM 403

fields the values of the two energy levels are nearly equal, both are nearly equally
populated, and the resultant magnetic moment is very small. In strong fiel~s. the
difference between the energy levels is large, the ordering effect predominates, and
nearly all magnets line up in the lower energy level where they have the same
direction as .1('.
It will be seen from Fig. 13-6that saturation, as predicted by quantum theory,
is very nearly attained when p 0 Jt'fkT - 3, or when

.Jt' = ~ .. 45 kOe K-1•

T fAD
Hence, if T - 300 K , a field of 13.5 x 10' Oe would be req uired for satura tion.
On the other hand, if the temperature is as low as I K, a field of 4.5 x 10' Oe
would produce saturation, and at a temperature ofO.I K, a field of only 4.5 x lOS
Oe would be required. (Modern superconducting electromagnets can produce
magnetic intensi ties up to 1.5 x 105 Oe.)
We now calculate the other thermodynamic properties of the system. The
total energy£, which in this case is the potential enecgy £ 0 , is

aIn Z1 )
E =E• .,. NkT (----ar- 1
- - Nk ( p - t anh fA
-n.Jt') -n.Jt'
-. (13- 37)
k kT
Comparison wi th Eq. (13-29) shows tha t the potential energy is

£ 0 = - .;('M. (13- 38)

The potential energy is negative because of our choice of reference level; that
is, the potential energy of a magnetic dipole is set equal to zero when the dipole is
at right angles to the field.
The heat capaci ty at constant .Jt' is

Cr - ( a~
ai'lr
• Nk ( fAnJ!'I.' sech•fAn.Jt'. ( 13-39)
krl kT
Figure 13-7 shows graphs of £ 1, and C1 , (both divided by Nk) as functions of
kTfp 0 .Jt'. The curves differ from the corresponding curves for the internal energy
and heat capacity of an assembly of harmonic oscillators because there are only
two permitted energy levels and the energy o f the subassembly cannot increase
indefinitely with increasing temperature.
 APPLICATIONS OF OUANTUM STATISTICS TO OTHER SYSTEMS 1:14

o.sr--T-"---r--.---,---,

0.4

0.)
c~ £
Nk iii<
0.2 -0.6

Fie. 13-7 The specific po1ential energy llnd specific

heat capacity at constant magnetic intensity, both
divided by Nk, for a paramagnetic cryslal as a
function of kT/1•n H·

Let us compare the heat capacity C1 of the magnetic ion subassembly with the
heat capacity c, o f the enti re crystal. LetT - I K and Jt'- 10' Oe. Then
kT J'nJt'
l'uJI' "" I.S, sech1 kT "" 0.81,
and by Eq. (13-39),
C1 ""Nk( I.S)- 1 x 0.81 "" 0.36 Nk.
Assuming there are SO nonmagnetic particles for every magnetic ion, and taking a
Debye temperature of 300 K as a typical value, we have from 1he Debye T' law,

Cv ""Nk(SO) x l~w' (3~)'.

"'o.s X 10- • Nk.
At this temperature, then, the heat capacity of the magnetic ions is about 100,000
times a.s great as the vibrational heat capacity of the crysral lauice. Much more
energy is required to orient the ionic magnets than to increase the vibrational energy
or the molecules or the lattice. It is this energy or orientation which allows the cooling
or the lattice during the process ofad iabalic demagnetization described in Section 8-8.
T he entropy of the subassembly can now be calculated from the equation
P - E- TS. From Eqs. ( 13-30) a nd ( 13-37) we have

S = -£ --- p.,..Jf) - -p.,.Jf'

F" = Nk [ In ( 2cosh-- P.nA'J
-tanh--. (13-40)
T kT kT kT
 13-5 NEGATIVE TEMPERATURES 405

t.O

o.n __ ___ .!!!.L _____ _

kT
......
Fie. 13-8 Th~ entropy of a paramagneti~
crystal.

Figure 13- 8 is a graph of S/Nk, plotted as a function of kT/p 11.YI'. AI a given

value of .:11', S approaches zero as T approaches zero. At this temperature, all
dipoles are in their to wer energy state; there is only one possible microstate; a nd
S = k tn 0 = k tn I = 0. At the other limit, when kT »
p 0 .YI',

cosh (p 0 .YI'fkT)- I, (p 0 Jt'/kT)- 0, tanh (p 11 Jt'fkT)- I,

and S - Nk In 2. The entropy is also a function of (Jt'/T) only. In a reversible

adiabatic demagnetization, S and hence (.;;f'/ T) remains constant. Thus as .:11'
is decreased, T must dec rease a lso in agreement with the thermodynamic result.

13- 5 NEGA TIV E TEMPERATURES

Consider again a system with just two possible magnetic energy levels, in which the
magnetic moment p 11 o f a particle can be either parallel or anti parallel to a mag-
netic intensity .;;f', The energy of the lower level, in which l'u is parallel to .Yi', is
<1 = - p 11.Yi'; and that of the upper level, in which p 11 is opposite to .:11', is
<2 = +p8 .YI'. I n the equilibrium state at a temperature T, the average occrpation
numbers of the levels are

n = -exp
I zN (-••)
-kT •

N
N, = :zexp ( -••)
kT .
 40t APPLICATIONS Of QUANTUM STATISTICS TO OTHER SYSTEMS 13-5

The ratio NJN, is

n,
-=exp - -
IJ, kT
(••- ''}
or

T- ![ ••- ,, J
k In IJ, - In IJ~'
(13-41)

and we can consider this as the equation defining T, in terms of<, ' " IJ, and IJ,.
If <2 > <1 and N, > IJ,, the right side of the equation is positive and Tis positive.
The situation ean be represented graphically as in Fig. 13-9(a), in which the
lengths of the heavy lines correspond to the average occupation numbers N, and
n,. I

N N

R,

« , - -p~

(1) (b)

Fig. 13-9 (a} In the state or stable equilibrium the occupation

number N, or the level or lower energy is larger than the
occupation number N, or the level or higher energy.
(b} Population inversion immediately after the magnetic
intensity Jt' has been reversed.

Now suppose the direction of the magnetic intensity is suddenly reversed.

Those magnetic moments that were parallel to the original field, and in the state
of lower energy <1 , are opposite to the new field and are now in the higher-energy
state, while those that were opposite to the original field, and in the higher·energy
state ••· are parallel to the new field and are now in the lower-energy state. Even-
tually, the moments in the higher-energy state will Oop over to the new lower-
energy state, but immediately after the field has been reversed, and before any
changes in occupation numbers have taken place, the situation will be that de-
picted in Fig. 13- 9(b). The average occupation number Ri of the new upper state
 13-6 THE ELECTRON GAS 407

is the same as the number IV, in the original lower state, and the occupation number
IVj of the new lower state is the same as the number IV, in the original upper state.
We say there has been a population inversion. Then if we consider that the tem-
perature of the system is defined by Eq. (13-41), and if T' is the temperature
corresponding to Fig. 13-9(b),

. 1[In •• - •.In J
T --
- k Nj - N~ .

Since IV; is greater than IVj, the denominator on the right side of the equation is
negative and T ' is negative.
Negative temperatures can be looked at from another viewpoint. At a tem-
perature T = 0, all magnets arc in their lower·energy states. As the temperature is
increased, more and more magnets move to the state of higher energy and when
T co + oo, both states are equally populated. Then one might say that if the
number in the higher state is eve~ gr.l!ater than that in the lower state, as it is when
there is a population inversion, the temperature must be holler than infinity.
We thus have the paradoxical result that a system at a negative temperature is
even hotter than at an infinite temperature.
In paramagnetic substances, the interactions between the ionic magnets and
the lattice are so great that the substance cannot exist in a state of population in-
version for an appreciable time. However, it was found by Pound, Purcell, and
Ramsey, in 1951, that the nuclear magnetic moments orthe lithium atoms in LiF
interact so slowly with the lattice that a time interval of several minutes is required
fo r equilibrium with the lattice to be attained, a time long enough for experiments
to be made showing that a population inversion actually existed.

13- 6 THE ELECTRON GAS

The most important example of an assembly obeying the Fermi-Dirac statistics

is that of the free electrons in a metallic conductor. We assume that each atom in
the crystal lattice parts with some (integral) number of its outer valence electrons
and that these electrons can move freely throughout the metal. There is, of course,
an electric field within the metal due to the positive ions and which varies widely
from point to point. On the average, however, the effect of this field cancels out
except at the surface of the metal where there is a strong localized field (or potential
barrier) that draws an electron back into the metal if it chances to make a small
excursion outside the surface. The free electrons arc therefore confined to the
interior of the metal in m~ch the same way that gas molecules are confined to the
interior of a container. We ~pea k of the electrons as an electron gas.
The degeneracies of the energy levels are the same as those of free particles
in a box, with one exception. There are two sets of electrons in a metal, identical
 408 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS 13-6

except that they have oppositely directed spins. The Pauli exclusion principle,
instead of asserting that there can be no more than one particle per state, now
permits t•·o electrons per state provided they have oppositely directed spins. This
is equivalent to doubling the number of sta tes in a macrolevel, or the degeneracy
fit§ of the macrolevel, and permitting only one electron per state. Hence, instead
of Eq. (12-17) we have
t. <§.,• s,m•v
~u
'l'l
v.
It will be more useful to express the degeneracy in terms of the kinetic energy
• - jmu'. Then since

• 2·
v - -
m
. flu = 21(;;;2)"'.-•11 l'l<,
it follows that

t. <§, - 4,y 2m)

(hi
111
r 111 l'l<. (13-4))
If for brevity we set

A e 4,y 2m) ,
(hi
111
(13-44)
then
l'l <§, • A•'" l'l«. (13-45)
The degeneracy therefore increases with the square root of the energy. Then from
the F-0 distribution function, Eq. (I J-40), the average number fl.AI' of electrons
in a macrolevel is
A% - A~. ~ A ,v• ~f. (13-46)
exp [(• - p)/kT) +I exp [(• - p)fkT) +I
The chemical potential p can be evaluated from the requirement that 2 1'lA" =
N, where N is the total number of electrons. Replacing the sum with an integral ,
we have
., . ,lit
N- A
L
• exp [(• - p)/kT) + 1
d<.

The integral cannot be evaluated in closed form and the result can be expressed
only as an infinite series. The result, first obtained by Sommerfeld, • is

[ "'(k9'
p • •r I · - - - - + ...J.
+ -"'(kT)' (13-47)
12 •r 80 •r
The quantity •r is a constant for a given metal and is called the Ftrmi mugy.
As we shall show, •r is a function of the number of electrons per unit volume, N/V,

• Arnold J. WISommerfeld, German physicist (1868- 1951).

 13-6 THE ELECTRON GAS 409

so the preceding equation expresses p. as a fu nction ofT and N/V. When T = 0,

p.0 = ••·· The distribution function at T = 0 is then

6%0 = 6 '§, (13-48)

exp [(• - <y)fkT] + I
The significance of the Fermi energy ••· can be seen as follows. I n all levels
~o r which < < ••·· the difference (• - .,.)is a negative quanti ty, and at T = 0,
~- E'F
--- = -00,
kT
The exponential term in Eq. (13-48) is then zero and in all levels for which
E < f'l-', 6A'0 = 6 '§, = A•'" 6<. {I 3-49)
That is, the ave rage number of electrons in a macrolevel equals the number of
states in the level, and all levels with energies less than • •· are fully occupied with
their quota of o ne electron in each state.

T,

Fll. 13- 10 Graphs of the distribution function of the

free electrons in a metal, at T = 0 and at two higher
temperatures T1 and T2 .

In all levels for which • > 'F• the term (< - <p) is positive. Hence at T = 0
the exponential term equals + oo a nd 6%" = 0. Thus there are no electrons in
these levels and the Ferm i e nergy <p is the maximum energy of an electron at
absolute zero. The corresponding level is called the Fermi level.
The solid curve in Fig. 13- 10 is a graph of the number of electrons per unit
energy interval, 6A~"0/6• = A•'"• at T = 0. The curve extends from • = 0 to
< = .,., and is zero a t all energies greater than <p.
An expression for the Fermi e nergy can now be obtained from the requirement
that 2 6%0 = N. Replacing the sum with an integral, introducing the d istri-
bution function at T = 0, and integrating over all levels from "zero to ••·· we have
 410 APPLICATIONS OF QUAN TUM STATISTICS TO OTHER SYSTEMS

or, after inserting the expression for A,

•F ~ ..!!:( 3N)tn. (13-50)

8m 1rV
Thus as stated earlier, •F is a function of the number of electrons per unit
volume, N/ V, but is independent of T.
As a numerical example, let the metal be silver, and since silver is monovalent we
assumd one free elec1ron per atom. The density of silver is 10.5 x I OS kg m->, its
atomic weight is 107, and the number of free electrons per cubic meter, N/ V, equals
the number of atoms per cubic meter which is 5.86 x 1020• The mass of an electron
is 9.11 x 1o->1 kg and h - 6.62 x l<r" J s. Then
<p - 9.1 X J0- 10 J - 5.6 cV.
The to tal energy U of the electrons is

u- ! •d.A", (13- 51)

or, replacing the sum with a n integra l,

U = AJ."' •"' d<.

o exp ((• - p.)/kT) +I
Again, the integral cannot be evaluated in closed form and must be expressed
as an infinite series. The result is

(13-52)
When T = 0,

(13- 53)

It is len as a problem to show t hat the same result is obtained if one inserts
in Eq. (13-51) the expression for the distributio n function at T- 0, a nd integra tes
from • = 0 to < = <y.
The mean energy per electron at absolute zero is
-._o = -U0 = -3 f .
N 5 F
Hence for silver,

;. = ~ x 5.6eV "' 3.5eV.

5
The mean kinetic energy of a gas molecule a t room temperature is only about
0.03 eV, and the temperature at which the mean kinetic energy of a gas molecule
is 3.5 eV is nearly 28,000 K. Hence the mean kinetic energy of the electrons in a
metal, e ven a t absolute zero, is much greater than that of the mo lecules of a n
ordinary gas at te mperat ures of many thousand kelvins.
 13-6 THE ELECTRO N GAS 411

At a temperature of 300 K, and for silver for which <r = 9.1 X 10-" J ,

kT = 1.38 X 10_,. X 300 = 4.58 ro-•.

X
•• 9.1 x to-"
Thus at this temperature the terms in powers of (kTf<,·),
in the series expansion in
Eq. (13-47), are all very small and to a good approximation one can consider that
p = •r at any temperature.
The dotted curves in Fig. 13-10 are graphs of the distribution function twY{!1<,
T
at higher temperatures 1 and T,, where T,
> T 1• It will be seen that the occupation
numbers change appreciably with increasing temperature, only in those levels near
the Fermi level. T he reason for this is the followi ng. Suppose the energy U of the
metal is gradually increased from its value U0 at T = 0, thus grad ually raising its
temperature. In order to accep t a small amount of energy, an electron must move
from its energy level at T = 0 to a level of slightly higher energy. But exc~t for
those electrons near the Fermi level, all states of higher energy are fully occupied
so that only those electrons near the Fermi level can move to a higher level when
the tempera ture is increased. With increasing temperature, those levels just below
the Fermi level become gradually depleted, electrons at still lower levels can move
to those that have been vacated, and so on.
For the particular level at which< = p, the quantity(< - p) ~ 0, a nd at a ny
temperatu re above T- 0, the exponential term in the distribution function equals
I, and the occupation number is
!1% = j!i '§,.
If the temperature is not too great, then to a good approximation p = <r
and to this approximation we can say that a t any temperature above T = 0, the
Fermi level is SO% occupied.
The heat capacity at constant volume, Cy, is given by

Cy - (:~t
and from Eq. ( 13-52),
"'(kT) [
Cy = - -
2 ••
Nk I - - -
10 ••
3"'(k9'
+ ··· . ] (13-54)

If the temperature is not too great, we can neglect terms in powers of (kT/•v)
higher than the fi rst, and to this approximation

"'(kT)
Cy = - - Nk. (13- 55)
2 ••
Replacing Nk with nR, where n is the number of moles, a nd dividing both
sides by n, we have for the molal specific heat capacity of the free electrons in a
metal,
c = -
"'(kT)
- R (13-56)
" 2 EJ> ,
 412 APPLICATIONS OF QUANTUM STATI STI CS TO OT HER SYSTEMS 13-6

which is zero at T = 0 and which increases linearly with the temperature T. For
silver at 300 K, using the value of (kTf•.-) previously calculated,
c. ~ 2.25 X 10-'R.
The molal specific heat capacity of a monatomic ideal gas, on the other hand,
is
3
c.~ lR.
Thus although the mean kinetic energy of the electrons in a metal is very much
larger than that of the molecules of an. ideal gas at the same temperature, the energy
changu only very slightly with changing temperature and the corresponding heat
capacity is extremely small. This result served to explain what had long been a
puzzle in the electron theory of metallic conduction. The observed molal specific
heat capacity of metallic conductors is not very different from that of noncon-
ductors , namely, according to the Dulong-Petit law, about 3R. But the free elec·
Irons, if they behaved like the molecules of an ideal gas, should make an additional
contribution of 3R/2 to the specific heat capacity, resulting in a value much larger
than that actually observed. The fact t hat only those electro ns having energies
near the Fermi level can incuast their energies as the temperatuJe is increased
leads to the result above, namely, that the electrons make only a negligible contri-
bution to the heat capacity.
To calculate the entropy of the electron gas, we make use of the fact that in a
reversible process at constant volume, the heat How into the gas when its tempera-
ture increases by dT is
dQ, = CydT = TdS;
and hence at a temperature T the entropy is

S =f.T dQ,T ~J.T CyT dT .

o o
Inserting the expression for Cv from Eq. (13-54) and carrying out the integration,
we obtain
"'(kT) [ "'(kT)' + · ··]
S = Nk-- I - - -
2 <v 10 •r
(1 3-57)

Hence the entropy is zero at T = 0, as it must be since there is only one

n
possible mi4rostate at T = 0 and at this temperature = I, S = kIn = 0. n
The Helmholtz function F is
F = U - TS,
and from the expressions derived above for U and S,

F- ~ N•,·[l _ 5,-'(kT)'+ . . ·] . (13-58)

5 12 ...
 PROBLEMS 413

T he p ressure P of the electron gas is given by

P= -(ot\
ovlr'
and since

it follows t hat

p = ~ N•r[t +
5 V
5"'(kT)'+ .. ·].
12 •r
(13- 59)

This is· the equation o f state of the electron gas, expressi ng P as a fu nction of V
and T.
Com pa rison with Eq. (13-52) shows that the pressure is two-th irds of the
e nergy density
p = ~!:!..
JV
Fo r sil ver, N/V "' 6 x tO'' elec trons per cubic meter and '• "" 10 x JO-lt J.
Hence at absolute zero,
P "' l x 6 x tO" x 10 x JO- lt "'24 x tO' N m- •
"" 240,000 atm!
In spite of Ihis tremendous pressure, the electrons do not all evaporate spontaneously
from the metal because of the potential barrier at its surface.

P ROBLEMS

13- 1 (a) Show that the entropy or an assembly or N Einstein oscillators is given by

S - 3Nk {exp (:;~ _ 1

- In [I - exp ( -9.,/T))}.

(b) Show that the entropy approaches zero as Tapproaches zero and (c) that the entropy
a pproaches 3Nk[ l + ln(T/O.,J) when Tis large. (d) Make a p lot or S/R versus T/92 .
13-2 (a) From Fig. 3-10 find the characteristic Einstein temperature 9;: for copper such
that the Einstein equation for c. agrees with experiment at a temperature of 200 K. (b)
Using this value of 8;:. calculate c. at 20 K and 1000 K and com pa re with the experimental
values. (c) Make a sketch oro., versus tempera ture so that the Einstein equation for c.
will yield the experimental values.
13- 3 The characteristic Debyetemperature for diamond is 1860 K and the characteristic
Einstein temperatu re is 1450 K. The experimental value or c. for diamond, at a tem-
pera ture of207 K, is 2.68 x 10' J kilomote-t K- 1• Calculate c. at207 K from the Ei nstein
and Debye equations and compare with experiment.
 414 APPLI CATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS

13-4 (a) Show that the heat capacity of a one-dimensional array of N coupled linear
oscillators is ~iven by
x11!~ dx •.
Cy • 3Nkr,.1 f.
(e• _ l )l'
0
where x - lr•/kT, and it is assumed that both longitudinal and transverse waves can
propagate along the array. (b) Evaluate this expression for Cv in the low and high
temperature limits.
13- 5 To show that the Debye specific heat capacity a t low tempera1ure can be deter-
mined from measurements of the velocity of sound, (a) show that

60 - ~(3N)'"
k 4,.v
where

..!. - !(~ + ~):

c' 3 c, c.
a nd (b) show that the specific heat per kilogram c, is
J6,.• k' r> r•
c, - Tiii,c'- 1.22 x 1o" pe''

where pis the density of the material. (c) Calculate the average value of the sound velocity
in copper. For copper , p is approximately 9000 kg m... and c, - 0.15 J kg-' K-• at s K.
(d) Calculate a value for 60 and for •,. for copper. (e) Calculate the val ue of Am•• and
compare to the in tera tomic spacing, assuming that copper has a cubic structure.
13-1 Calculate values (a) for c1 and c1 of Eq. (13- 22) and (b) for the Stefa n-Boltzmann
constant"·
13-7 (a) Show that for electromagnetic radiation the energy per unit volume in the wave-
length range between A and A + dA is given by
S..lrc dA
du, • A' exp (lrc/ Akn - I ·
(b) Show that the value of A for which 6 u, is a maximum is given by A,.T - 2.9 x 1o-•
m K . This is known as Wien's displacement Jaw. (c) Calcula te A,. for the earth, assuming
the earth to be a blackbody.
13-8 (a) Show that Wien's Jaw can be derived by assuming that photons obey M-B
statistics. (b) Show that Wien's law results in a total energy densi ty which is nearly the
same as that derived in Section 13- 3.
13-9 If the magnetic moment I' • Jl'n of an atom is large enough, there will be 2J + I
possible angles 6 between the magnetic moment a nd the applied magnetic intensity Jf'
corresponding 10 magnetic levels having energies ~J • mJpl where nrJ has values
between -J aod + J. (a) Show that Z1 will be given by
. h (2J + I)I'Jt'
2kT
z- Sin
_ _ .::..:_ ( 1 3~)
. I'Jt'
smh 2/cT
 PROBLEMS 4tl

[Hint: Sec the derivation of Eq. (12-44).) (b) Show that the net magnetic moment of the
system is given by
(2J + I)
M • Np. [ - --
p.Jl'
coth (2J + I) 2kT - coth 2kf .
p.Jl']
2

Th is is called the Brillouin• function. (c) Show that the net magnetic moment follows
Curie's law in the limit of high temperatures and low fields. (d) In the limit of low tem-
perature and high fields, show that all the dipoles are aligned. (e) Show that the expression
for the net magnetic moment derived in part (b) reduces to Eq. (13- 29) when 2J + I • 2
andg- 2.
13- 10 Use Eq. (13- 60) of the previous problem to calculate the entropy o f N distin-
guishable magnetic dipoles. Evaluate the expression in the limit of high and low tem-
peratures and make a graph or the entropy as a function ofT and .II'.
13-11 A paramagnetic salt contains 10" magnetic ions per cubic meter, each with a
magnetic moment of I Boh r magneton. Calculate the difference between the number of
ions aligned parallel to the applied intensity of 10 kOe and that aligned anti parallel at
(a) 300 K, (b)4 K, if the volume of the sample is 100 em•. Calculate the magnetic moment
or the sample at these two temperatures.
13-12 Use the statistical definitions or work, total energy, and net magnetic moment
to show that the work of magnetization is givtn by dW - -Jl'dM. (Hint: See Section
3- 13.1 1
13-13 Derive expressions for the magnetic contribution to the entropy and the heat
capacity at constant magnetic intensity .II' for the system d iscussed in Section 13-4. Sketch
curves of these properties as a funct ion of .11'/T.
13-14 Calculate the mean spe<d, the root-mean-square speed and the mean-reciprocal
speed in terms of u F • (2<~nr)''' for an electron gas at 0 K.
13-15 (a) Show that the average number or electrons having speeds between • and
u + du is given by
lwY _ 8nm'JI u' Av
• lr' exp [(jmu' - p.)JkT] + I ·
(b) Sketch Ao~YJ Au as a function or u' at T - 0 K.
13-16 (a) Calculate '•· for aluminum assuming 3 electrons per aluminum atom. (b)
Show that for aluminum at 1000 K , 1• difTers from ,,. by less than 0.01%. (c) Calculate
the electronic contribution to the molal specific heat capacity or aluminum at room
temperature and compare it to JR. (The density or aluminum is 2.7 x 10' kg m-• and
its atomic weight is 27.)
13-17 The Fermi velocity is defined as Vp • (2<y/m)111 and tht Fermi temperature as
Tv • ' F/k. (a) Calculate values of the Fermi velocity, momentum, and temperature for
electrons in silver. (b) Determine the magnitude of the second term in Eqs. (13-47),
(13- 52), (13- 54), ( 13- n}, (13-58), and (13- S9) at room temperature. (c) At what ttm·
pera ture does the second term contribute approximately a I % correction in the above
equations?

• Leon N. Brillouin, French physicist (1889-1969).

418 APPLICATIONS OF QUANTUM STATISTICS TO OTHER SYSTEMS

JJ-18 F ind the mean energy per electron by substituting the expression for ll./V' into
Eq. (13-51).
I J-19 Derive Eqs. (13-57), (1 3-58), and (13-59).

I J-ZO In a one-dimensional electron gas tJ.<§, - Tv'2mf< /J., where L is the length of

h'N'
the sample of N electrons, (a) Sketch .A'"O(•) as a function of'· (b)Show that <y • mL' .
32
(c) Find the aver11ge energy per electron at 0 K.
I J-Z1 (a) Use the data shown in Fig. 7-7 to determine the Fermi energy of liquid He'
which can also be considered as a gu of particles obeying Fermi-Dirac statistics. (b)
Determine the Fermi velocity and temper11ture for He'. (See Problem 13- 17).
13-22 The free electrons in silver can be considered an electron gas. Calculate the com-
pressibility and expansivity of this gas and compare them to the experimental values for
silver of 0.99 x 10- 11 m' N-1 and 56.1 x Jo-t K-1 , respectively.
 Appendix
A SELECTED DIFFERENTIALS FROM A CONDENSED COLLECTION OF
THERMODYNAM IC FORMULAS

B THE LAGRANGE METHOD OF UNDETERMINED MULTIPLIERS

C PROPERTIES OF FACTORIALS

0 AN ALTERNATIVE DERIVATIO N OF DISTRIBUTION FUNCTIONS

MAGNETIC POTENTIAL ENERGY

 A
Selected differentials from a condensed
collection of thermodynamic formulas
by P. W. Bridgman

Any par1 ial derivative of a state variable of a thermodyna m ic system, with re spect
to any other state vuriable, a third variable being held constant [for exam ple,
(oufov)rJ can be written, from Eq. (4-20), in the for m

~ I~
(vuvu)r- (oufoz)r
---
(oufoz)r
where z is any arbitrary state fu nc tion. Then if one ta b ulates the partial ae rivatives
o f all state variables with respect to an a rbitrary function z, any partial derivative
can be obtained by dividing one tabulated quantity by anot her. For brevity,
d erivatives of the form (oufoz)r are wri tten in the table below in t he symbolic
fo rm ( ou)r. Then, for example,

+ P(oufoP)r =
( ou r
~) _ (ou)r _ T(oufoT)p TP _ P
(ou)r -(oufoP)r 1e '

which agrees with Eq. (6-9). Ratios (not derivatives) such as d'q,./dup can be
treated in the same way. For a further discussion, see A Condtnud Coi/J:t/on of
Thumodynamics Formulas by P. W. Bridgman ( Ha rvard University Press, 1925),
from which t he table belo w is ta ken.

P constan t T constant
can,.- 1 (iJP) r - -1
(ou),. = (ovfoT),. (ov)r = -(oufoP)r
(os),. = cp/T (os)r ~ (iJvfoT)p
(oq)p- ,,. (oq)r - T(ou/oT),
(ow)!' = P(oufoT)ro (ow)r - -p(ou/oP)r
(ou),. = c,. - P(iJufoT)p (ou)r = T(ouf oT),. + P(oufoP)r
(iJh)p = Cp (oh)r - -• + T (ovfoT),.
(og)ro = -s (og)r - -v
(of),. = - s -P(iJu/iJT)p (iJf)r - P(oufoP)r
419
 420 APPE NDIX A

h constant g constant
(oP).- - c1 , (oP>, = s
(or). - v - T(oufoTlJ. cor>, - .
(o•>• - - cp(ou/oP)r - T(oufoT) 1~ (ou), - u(oufoT),, + s(oufoP)r
+ u(oufoT),,
(os), = .!. [vcp - sT(oufoTlrl
T
(oq).- vcr (oq), = -sT(oufoTlr + ucr
(o w). = -P[cp{oufoP)r + T(oufoT)~ (ow), = P[u(oufoT )" + s(oufoP)rl
- u(oufoTlJ.J

~ u constant
(oP), - - cpfT (oP). - -(oufoT)p
(oT ), - - CoufoT)p (oT). - (oufoP)r

( ou), - - !T [cp{oufoP) r + T (ouf oT)J.J (os). =!T [c1i}ufoP)r + T(oufoTll ·l

(oq), - o (oq). = c1.(ovfoP)r + T(oofoT)J.
(ow). = o
(ow), - - !. [cp{oufoP)r + T(ou(OT)~J (ou). = c1.(oufoP)r + T (oufoT)j.
rr
p (olr). = c1.(oufoP)r + T (oufoT)J,
(ou), - T [cr(oufoP)r + ( T oufoT)J.J - u(oufoT), ,
(oh), - - vcp/T (og). = -u(ovfoT) 1• - s(ovfoP)r
(of). ~ -s(oufoPlf·
( og), - - ! [ucP - sT(oufoT), .J
T

(Of), = ! [Pcp(oufoP)r + PT(oufoT)j,

T
 B

The Lagrange method of undetermined

multipliers

In an algebraic equation such as

ax+ by= 0, (B- 1)
one is accustomed to consider o ne of the variables, say x , as the inclepenclent
variablo and the other variable, y, as the clepenclent variable. The equation is then
considered as imposing a relation between the dependent and independent variables
in terms of the coefficients a and b, namely, in this case, y = -(afb)x.
Suppose, however, that both x andy are independent variables. Then y may
have any value regardless of the value of x, and we can no longer require that
y = - (a/b)x. The equation ax + by = 0 can be satisfied for any pair of variables
x and y only if a = 0, b = 0.
Suppose next that x and y are not completely independent but must also
satisfy a condition equation, which we take, for example, as
X+ 2y = 0. (B-2)
What can we now say about the coefficients a and bin Eq. (B-1)? One p rocedure
is to consider Eq. (B- l) and the condition equation (B-2) as a pair of simultaneous
linear equations. We solve Eq. (B-2) for x and substitute in Eq. (B- 1):
X= - 2y
a(-2y) +by= 0,
b = 2a. (B-3)
Then Eq. (B- 1) is satisfied for any pair of values of a and b that satisfy Eq. (B-3),
provided the values of x andy satisfy the condition equation (B-2).
If the number of independent variables and condition equations is small, the
procedure above is adequate. But when these numbers become very large, there
are too many simultaneous equations to solve. In this case, we use the Lagrange•
method of undetermined multipliers. Each condition equation is multiplied by an
undetermined constant A. If there are k condition equations, there are k such

• Joseph L. Lagrange, French mathematician (1736-1813).

421
 422 APPENDIX 8

multiplters: l, A,. ... , A•. In our problem there is only one such equation and
one multiplier A. Then from Eq. (B-2),
.lx + 2Ay = 0. (B-4)
Now add this to Eq. (B-1), giving
(a + A)x + (2A + b)y = 0. (B-S)
Now assign a value to Asuch that the coefficient of either x or y is zero. If we choose
x, then (a + A) = 0; A = -a. (B-6)
Equation (B-5) then reduces to
(2A + b)y = 0, (B-7)
which contains only one of the variables. But since either one of the variables can
be considered independent, Eq. (B-7) is satisfied only if
(2A +b) = 0; b ~ -2A. (B-8)
Then from Eqs. (B-6) and (B-8) we have
b = 2a, (B- 9)
which is the same~ Eq. (B-3).
In effect, the usc of Lagrange multipliers leads to an equation, Eq. (8-S),
which has the same property as if both x andy were independent, since the coeffi-
cient of each is zero.
We now use the Lagrange method of undetermined multipliers to explain how
Eqs. (8-29), the equations of phase equilibrium, are a necessary consequence of
Eq. (8-27), which expresses the condition that the Gibbs function shall be a
minimum, subject to the condition equatio ns (8- 28). If the values of the dn\11's
in Eq. (8-27) were completely independent, the equation could be satisfied for an
arbitrary set of the dn\11 's only if the coefficient of each were zero. The method of
undetermined multipliers takes the condition equations inti) account so as to
eliminate some o f the terms in Eq. (8-27) to obtain an equation in which the re-
maining dn\1hs are independent, so that the coefficient of each can be set equal to
zero. The procedure is as follows.
We multiply the first of the condition equations (8- 28) by a constant .l1 whose
value for the present is undetermined. The second equation is multi plied by a
second constant l,, the next by a constant .l3 , and so on. These equations are then
added to Eq. (8-27). T he result is the equation

<.uP' + l,) dn: 0 + (p:" +. A1) dn:'' + · · · + (p:•' + .l1) dn:•l

+ <.u:" + A1) dnl'' + (.ul'' + A,) dn~" + · · · + (!4'' + .l,) dn~·~

-':J.'
 APPEND IX B 423

The total number of dnl11 's in this equation is kTT, one for each of the k con-
stituents in each of the TT phases. For any constituent i, arbitrary values may be
assigned to the dn,'s in all phases but one, ma ki ng a total of(,. - I) arbitrary
values. The remaining dn 1 then takes up the slack, since

·-· 11
Idn l - 0.
I•I
Then since there are k constituents, the total number of dnl11's which can be given
arbitrary values, or the number that are independent, is k(,. - I) - k,. - k .
Let us therefo re assign values to the (as yet) undetermi ned multipliers, such that
for each constituent I, in some one o f the phases j, the sum (,ul11 + .11) - 0. For
example, let us select phase I and assign a value to 11 such that in phase I

(,u\11 + l,) = 0, or ,u\11 = -l,.

Then the product (,u\11 + 11) dn\11
is zero regardless of the value of dn\11
and this term drops out of the sum in Eq. (B- 10). In the same way, we let

{.ul11 + l,)- 0, or ,ul11 = - .1,,

a nd so on for each of the k constituents. This reduces the number of dnl11's in
Eq. (B-1 ) by k, leaving a total of hr - k. But since this is the number of dnl11 's
that can be considered independent, it follows tha t the coefficient of each of the
remaining dnl11's must be zero. Therefore for any constituent i in all)' phase j,
,u:IJ a -A,.
Therefore the c hemical potential of any constituent i has the same value - 11
in all phases, which leads to the equations of phase equilibrium, Eqs. (8-29).
N ote that the values of the .lo's themselves need not be known ; the only significant
aspect is that the values of the chemical potentials of every phase are equal,
whatever these values may be.
One can consider that, in t./fut, the method of Lagrange multipliers makes a//
the dnl 11's in Eq. ( B- 10) independent, since the coefficient of each is zero , but the
coefficients are zero for different reasons. In phase I, the coefficients ~re zero
because we assigned values to the .l's to make them zero. In the other phases, the
coefficients a re zero because the remaining dnl" 's are independent.
The choice of phase I in the preceding argument was not essential; we could
equally well have started with any other phase a nd, in fact, could have selec ted
different phases for each constituent. In any case, we would eliminate the same
number k of dnl11's from Eq. (B-10), and the remainder would be independe nt.
 c
Properties of factorials

In the derivations of the distribution functions of particles obeying the vario us

statistics, many properties of the factoria l are used. In this a ppendix we derive
these properties by investigating the gamma function r(s),. Stirling's approximation
for calculating factorials of large numbers is also developed.
The factorial of a positive integer n is written nl and defined as
n! = n(n - l )(n - 2) · · · I. (C-1)
From this ~efi n ition it follows that
(n + I)! ~ (n + l)n!. (C- 2)
Equation (C-2) can be used to determine 0 ! and ( -n)!
If n - 0, Eq. (C- 2) gives I! = (01) and
0! = I. (C-3)
l fn =-I , Eq. (C-2) resul ts in the expression 0! = 0(-1)!. Since 0! =I ,
we can take (-I)! to be oo, tha t is
(-I)!= oo. (C-4)
Ho wever, this involves di vision by zero which is undefined mathematically. The
gamma function is a n expression for val ues of n which may not be integer,
which yields Eqs. (C- 1) to (C- 3) for integer n. In the limit that n a pproaches -I,
the gamma function approaches oo.
Integrals of the form
f(s) = [ a(t)e-•' dt

are called Laplace• transforms. They are ve ry useful in many branches of science
and engineering. The gamma function is a Laplace tra nsform in which s - I and
a( I ) - I" where n need not be an integer. Thus

/(I ) - r(n + 1) =[ t"e-• dl. (C- 5)

• Marquis de Pierre S. Laplace, French mathematician (1749- 1827).

424
 APPENDIX C 425

For n ~ -I, integration by parts yields

f.~ t •e-• dt = -t•e-• [ + n f.~ r•-'e-• dt.

T he first term on the right is zero at both limits since .-• approaches zero faste-
than r• approaches infinity at the upper limit. Then

J.~ r•e-•dt = n[ ,•-•e-• dt

or
r (n + I) - nr (n). (C-6)
The gamma function can be successively integrated by parts so that

r (n + I) = n(n - l)(n - 2) · · · I,
and if n is an integer
r(n +I) = n!. (C-7)
If n = 0, the gamma function can be integrated directly and

r(J) = J.~ a.-• = 1.

Since by Eq. (C-7), r(J) = 0!,
0! =I , (C- 8)
in agreement with Eq. (C-3) .
The integral of Eq. (C- 5) diverges if n ~ -I , but by rewriting Eq. (C-6) as

(C-9)
the definition of r(n) can be extended to negative integers. If 0 < n < I, r(n)
can be determined from Eq. (C-9). Using this recursion formula again, the values
of r (n) for - I < n < 0 can be found from the values for r (n) when 0 < n < I,
and so on. Thus r(n) is determined for all non integer values of n.
However, since r (l) = I the method fails for n = 0, since di vision by zero
is undefined. Thus
lim r(n) =lim n- 'r(n + I) = ±00. (C- 10)
ft-+0 ft-+ 0

Similar behavior is found for all negative integers.

For small values of n the factorial can be eval uated by direct computation.
H owever, it is often necessary to evaluate n! for large values of n. The factoria l
of a large num ber can be found with sufficient precision by Stirling's approxima-
tion which we now derive.
The natural logarithm of factorial n is

In (n!) = In 2 + In 3 + · · · + Inn.
 426 APPENDIX C

Fig. C-1 A graph of Inn as a function

of n.

This is exactly equal to the area under the step curve shown by dotted lines
in Fig. C-1, between n - I and n - n, since each rectangle is of unit width and
the height of the first is In 2, that of the second is In 3, etc. This area is approxi-
mately equal to the area under the smooth curve y - In n between the same limits,
provided n is large. For small values of n the step curve differs appreciably from
the smooth curve, but the latter becomes more and more nearly horizontal as n
increases. Hence approximately, for large n,

In (n !) = J"ln n dn.
Integration by parts gives
In (n!)- n ln n - n + I,
and if n is large we may neglect the I, so finally
In (n !) - n In n - n. (C-11)
This is Stirling's approximation.
An exact analysis leads to the following infinite series.

n! = J2,m {!!)"[I + ...!.. + _I_ - _ill_ + .. ·J. (C- 12)

e 12n 288n' 5 l840n'
If all terms in the series except the first are neglected, we obtain
In (n !) = i In 2" + i In n + n In n - n. (C-13)
If n is very large compared with unity, the first two terms of this expression are
negligible also, and we get Eq. (C-11).
 D

An alternative derivation of distribution

functions

At the end of Section 11-5, it was· noted that when the number of particles in an
assembly becomes large, the occupation numbers of the levels in the most probable
macrostate are very nearly the same as the average occupation numbers for the
assembly. This is not only true for particles obeying B-E statistics, but it holds
equally well for the other statistics. Thus when the system is in equilibrium, the
distribution of particles among levels can also be determined from t he occupation
numbers of the macrostate with the maximum thermodynamic pro bability, sub-
ject to the constraints that the total energy and the total number of panicles of the
assembly is constant.
When one looks at a large number of identical assemblies, one macrostate
occurs the most often. The assumption is that this macrostate is the distribution of
particles among levels for the system in equilibrium. Therefore the properties of
t he system are determined by the distribution of particles among levels that has
the maximum thermodynamic probability. In the text we assume that the prop-
erties of the system are determined by the average occupation numbers of the
levels. In the limits of large numbers of particles both methods lead to the same
distribution functions, as we shall show.
We now describe the conventional procedure for calculating occupation
numbers in the most probable macrostate, or, the most probable occupation
numbers. If we let ?f'"• represent the thermodynamic probability of the most
probable macrostate, the entropy S is set proportional to the logarithm of "If'"•,
that is,
S = k8 1n 'If'"•.

To find the most probable macrostate, we use the usual criterion for thel maxi-
mum value of a function, namely, that its first variation is equal to zero. (Strictly
speaking, it should also be shown that this leads to a maximum value and not to a
minimum.) We shall illustrate by considering the Maxwell-Boltzmann statistics,
although the same procedure can be followed in the other statistics as well.
427
 428 APPENDIX D

In the M-B statistics, the thermodynamic probability of a macrostate is given

by Eq. (I 1-21),
gNI
irM- 8- N! IT1 ...L
N !
(0-1)
1

Instead of maximizing ill", it is simpler to maximize In if/', since if if/' is a

maximum, its logarithm is a maximum also. Then considering the thermo-
dynamic probability of the most probable macrostate,

In -tr• = InN! + L N1 In g1 - L In N1 !. (0-2)

I I
We assume that N)) I, and that in any level j, N1 )) I, so that we can use the
Stirling apvroximation (see Appendix C), and

InN! - N InN- N,
In N 1 ! - N 1 1n N 1 - N 1•
Then
In -tr• = N InN- N + I N 1 In g1 - I N 1 In N 1 + I N 1•
I I I
But I N1 = N, so

In -tr• = N InN + I N 1 In g1 - I N 1 In N 1 - N In N - I N 1 In.!!.. (D-3)

I I I Nl

Now compare this macrostate with a neighboring macrostate in which the

occupation numbers are slightly different. Let the occupation number of any level
j differ from its most probable value by ~N1 • Since ~N1 N1 , we can use the «
methods of differential calculus, considering ~N1 as a mathematical differential .
T he differential of In tr• is then, since Nand g1 are constants,
6 In tr• = I In g1 6N1 - L 6N1 - L In N1 6N1• (0-4)
I I I
Since the total number of particles is the same in the two macrostates, any
increases in the occupation numbers of some levels must be balanced by dec reases
in the occupation numbers of other levels: and hence I 1 6N1 - 0. Since In ill"*
is to be a maximum, we set 6 In -tr• - 0. Then

or

(O- S)

If the N/s were indtptndtnt, then as explained in Appendix B, this equation

could be satisfied only if the coefficient of each 6N1 were zero. But the 6N,'s arc
 APPENDIX D 4~

not independent. We have shown above that

6N = 2 6N1 - 0; (D-6)
I
and since the total energy U - 2 1 <1N 1 is the same in both macrostates, any in-
crease in energy resulting from an increase in the occupation number of a level
must be balanced by a decrease in the energy of other levels and a second condition
equation is
6U = 2•1 6N1 =0 (D-7)
I
We therefore use the Lagrange method of undetermined multipliers described
in Appendix B. Multiply the first condition equation, Eq. (D-6), by a constant
which for later convenience we write as In "• multiply the second by a constant
- p, and add these products to Eq. {0-5), obtaining

~ (In*.+ In«- P•;) 6N 1 = o.

In effect, the 6N,'s are now independent and the coefficient of each must be
zero. Hence for any level j,

In.!! + In" - P•1 • 0, (D-8)

N,
or
N1 = «g1 exp(- P<1) , (D-9)
which is the distribution function for the most probable occupation numbers,
expressed in terms of the constants " and p.
Now sum the" preceding equation over allj's, and let
Z = 2 g1 exp(- P<1)
I
where Z is the single particle partition function described in Section 11- 14. Then
since 21 N1 • N, it follows that
N
a. :: z· (0-10)
and from Eq. (D-9),
N1 N
- = - exp( - P•1). {D-11)
g, z
To evaluate the constant p, we insert in Eq. (D-3) the expression for In (g;/N1)
from Eq. (D-11), and setS - ku In tr•, giving

S = ku[N InN- 2J N 1 InN + 2 N1 In Z + p 2 •1N1] ,

J J
or
S = Nku In Z + Pk0 U. (D-12)
 430 APPENDIX 0

If the energy levels are functions of the volume V (or some other extensive
parameter), then Z is a function of Pand V and has the same value in two equi-
librium states in which the values of Pa nd V are the same. The entropy difference
S between the states, since In Z is a constant, is

t:.S - #o t:.U. (0-JJ)

From the principles of thermodynamics, the entropy d ifference between two
equilibrium states at the same tem perature and volume is

It follows that pk 8 - l{T, or

P-=-'- . (0-14)
k8 T
Hence Eq. (0-12) can be written

S- Nk 8 lnZ + Y..,
T
(D-15)
and
F :aU- TS =- Nk 0 T inZ. (D-16)
The chemical potential f' is

f' - (~F) =- -k 0 T In Z, (D-17)

~N T.Y
and hence
-I' I f'
lnZ =- - , - = exp--. (D-18)
k8T Z k0 T
The distribution function, from Eq. ( 0 - 11), can now be written as

!!..!. "" N exp f' - '•. ( D-19)

g1 k0 T
Comparison with Eq. ( I 1-44) shows that the distribution function for the most
probable occupation numbers is given by the same expression as that for the
arNrage occupation numbers.
One objection to the conventional procedure is that if a n N1 is calculated from
the preceding equation, the value obtained is not necessarily an inleger, while the
actual occupation number of a level is necessarily integral. If we consider the right
side of Eq. (0-19) does give the correct values of the average occupation numbers,
this equation can be interpreted to mean that the occupation numbers in the most
probable macrostate are the nearest integer to their values averaged over all
macrostatesi Since the occupation numbers are all very large, the "nearest integer"
will differ by only a ulalively small amount from the average.
 APPENDIX 0 431

A more serious objection is the following. One of the terms in the expression
for the thermodynamic probability of a macrosta te in the Fermi-Dirac statistics
is (g1 - N1) !. If In (g1 - N1)! is evaluated by the Stirling approximation, a nd the
procedure above is followed, one does obtain the same expression for the most
probable occupation numbers as that for their average values. But in the F·D
statistics, the difference (g1 - N1) is not necessarily a large nu mber a nd may in
fact be zero if a level is fully occupied. The use of the Stirling approximation to
evaluate In (g1 - N1)1 is therefore questionable, even if it leads to the right answer.
The procedure followed in Section 11- 10, however, does not require the use of
Stirling's approximation and is valid provided only that the N,'s themselves arc
large numbers.
 E
Magnetic potential energy

Each magnetic ion in a paramagnetic crystal is a small permanent magnet and is

equivalent to a tiny current loop as in Fig. E-l . The ion has a magnetic moment
p., which if the ion actually did consist of a current I in a loop of a rea A, would
equal (in the system of units we arc using) the product /A. The moment can be
represented by a vector perpendicular to the plane of the loop.
If the moment vector makes an angle 9 with the direction of an exteroal
magnetic field of intensity .;t', a torque T of magnitude JJ.;t' sin 0 is exerted on the
loop, in such a direction as to align the magnetic moment in the same direction as
.;t'. I n Fig. E-l, this torque is clockwise. In the usual sign convention, the angle
8 is considered positive when measured counterclockl<'ise from the direction of 9,
so we should write
T - -p.;tt' sin 8. (E-l )
If the loop is given a small counterclockwise displacement, so that the angle 8
increases by dO, the work of this torque is
dW = T dO - -p..;t' sin 0 dO.
The increase in magnetic potential energy o f the loop, dt 0 , is defined as the
negati~ of this work, just as the increase in gravitational potential energy of a
body of mass m, when it is lifted vertically in a gravitational field of intensity g,
is the negative of the work of the downward gravitational force -mg exerted o n it.
Hence
d• 0 - p.;tt' sin 9 d8. (E-2)
The total change in potential energy when the angle 9 is increased from 91
to 9, is
..
. . f.'•
'• - '• - p. Jt' sin 8 dO = JW('(cos 6 1 - cos OJ.

Let us take the reference level of potential energy as that at which the moment
is at right angles to the field, where 0 = 90• and cos 0 = 0. Hence if we set
432
 APPENDIX E 433

Fig. E- 1 A magneuc 10n of magnelic

moment I' is equivalent 10 a small
current loop.

81 - 90° and '•• = 0, and let c1,, a nd 01 refer to any a r bitrary angle 0,
c0 - 0 - p.Jt'(O - cos 8),
and
•• = -p.Jt' cos 0. (E- 3)
When t he angle 8 is less than 90°, as in Fig. E- 1, cos 0 is positive and the
potential energy •., is negative. That is, the po tential energy is less than tha t in the
reference level. When 0 is greater than 90°, cos 8 is negative and •. is positive.
Let ll.A~", be the number of a tomic magnets whose moments make angles with
t he field between 6 and 6 + 68. Each of these has a component mom~nt in the
direction of the field of p. cos 0, and the moment d ue to these is
6M = fl.AI', p. cos 8.
The total moment M of the entire crystal is
M - 1: fl.AI', p. cos 8. (E-4)
In the same way, the total potentia l energy E,, of the crysta l is
E" = -1: liJV', p.Jf' cos 8.
It follows from the two preceding equations that
E.= - Jt'M. (E- 5)
I
f
i>

I. I·
C!
I·

'
1-
0.
!
I·
1-
1-
1-
1-
1-
1-
1-
1-
1-
de

1-

(b:

{c)

I-
I-
1-
1-:
{a<
pre
Ch
2-:
2-:
I
I Answers to Problems
Chapter 1
1-1 (a) no; (d) yes.
1-l (a) excensive; (d) incensivc.
1- 3 (a) JO>kgm->; (b) to->m1 kg-1; (c) 18 x to-'m'kilomole-1; (d) 1.29kgm- •,
o.n5 m' kg-'. 22.4 m' ki lomole-1.
1-4 Abouc tOO Torr.
1-5 (b) I.Ot x tO' N m- •.
1-6 (a) 4.
1-7 (c) decrease.
1-8 t B K, t85 K, t93 K, t97 K.
1- 9 (a) 328 K; (b) 6.84 em; (c) no.
1-10 (a) a - 1.55 x to-•, b - -115; (b) t l2 degrees; (c)5.97 em.
1- 11 (a) 73.3; (b) 26.7 degrees.
I-ll (a) 672; (b) t80 degrees.
1-13 (a) A - 3.66 x t O-' ai m K- 1, B - 321 degrees, C- 3.66 x to-> K- 1; (b) t30
degrees; (c) O. t 2 arm; (d) - «>.

(b) a - 2.5 degrees m v-1, b - 0;

(c)

l - ts (a) -t95.so•c; (b) t 39.23 R; (c) -320.44°F.

1- 16 (a) t 4.20 kelvins; (b) t4.20 deg C; (c) 25 .56 rankines; (d) 25.56 deg F.
1-17 (a) no; (b) yes.
1- 21 (a) reve~>ible isobaric process; (b) q uasiscacic isochermat process; (c) irreve,.ible
(adiabacic) compression; (d) irrevmible isochoric process; (e) reversible isochermal
process; (f) irrevmibte adiabacic process.
Chapter 2
Z- 2 (a) 5.7 x to-1 m• kilomolc-1; (b) 8.8 kilomoles; (c) 5.3 kilomolcs.
2- 3 (a) A - Pl/RT1 ; (c) 800 K.
2-4 (a) 0.25 m; (b) 500 Torr.
435
 431 ANSWERS TO PROBLEMS

l-5 (a) ~56 K.

%- 6 0.18 m.
%-7 8.66d.
%-9 (a) 300 K; (b) 6.24 m' kilomolc-•; (c) 150 K , 120 K; (d) 10m'; (c) 8 kg.
%-10 (a) 0.308 kilomoles ; (b) 9.86 kg; (c) 3.96 x 10" N m-•; (d) 0.277 kilomoles.
%-11 (a) I m'; (b) ISO K ; (c) 200 K, 0.67 m'; (d) 22S K, 0.75 m'.
l -13 (b) 0.06, 0.22, 0.51.
l - 14 (a) 4.87 x 107 N m- •; (b) 5.10 x 107 N m-•; (c) 8.31 x 10' and 8.70 x 10' 1 kilo·
molc- •K- 1•
l-19 6.S x 101 N m-•.
%-:!3 (a) fJ • (u - b)/uT, K • (u - b)'/RTu.
2-25 v • •o cxp (aT'/P), a/b • l.
l-26 (a) f.oa.; (b) L,J.YA)-1 ; (c) -l!.FfoYA.
2-27 (a) 2.88 x 101 N ; (b) 6 m.
l-29 (a) 0.031 m' kilomolc-•; (b) 0.042 m' kilomolc- •.
2-30 (b) 0.270.
l-32 [(• - b)(uRT + a)JIT(a(u - b) - ,SRn.
l-33 (a) Rf(u - b); (b) R/(• -b); (c) (exp ( -a/uRT)X• - b)-1(R + afvT).
l-35 (b) 10""1'(6.4 + 3.3 X lo-'T)m' N-1; (c) - 3.3 X 10- 11 m 1 N-1 K - 1;
(d) S.2 X 10""1•

Ch•pter' 3
3-1 1.69 x 10° 1.
3-2 1.91 X I()> 1.
3-3 - 3nRTJ8.
3-4 2.03 J.
3-5 1.13 1.
3-6 (b) Work on the gas; (c) 8.1S x 10' 1, 0.4341; (d) 0.4 m 0, 1.44 x Ia-' m'.
3-7 (a) W • RTin ((v1 - b)/(u 1 - b)] + a((l/v1 ) - (l /v 1)]; (b) 4.26 x 10' J;
(c) 4.3 x 10' J.
3-8 (b) d'W • nRdT + nRTdP/P.
3-9 (a) d'W • - Ft..(dF/YA+ «dT); (b)W,. •-Ff.oa.(T1 -T1); (c)WT •
- 4(FI- Ff>/2 YA.
3-10 (a) d'W • -Cc-J{'dJt'/T + Cc-Jf"dTfT'; (b)W_... • -CL..;t''(lfT, - I/T1) ;
(c) w., • -(Ccf2T)(.JY; - Jt"l).
3-JJ - 3(/10 V + C0 /TJJt';/2.
3-13 -2.03 X 10° J.
3-14 (a) -3.11 x 10' 1; (b) -4.32 x 10' J ; (c) I SO K; (d) 1.2S x I()> N m-•.
~16 W, • 0, W, • 11.2 X J()' J, W, • -8.08 X J()' 1, Wok o • 3.12 X JO' J.
 ANSWERS TO PROBLEMS 437

3- 17 (a) 6 x 101 J ; (b) clockwise.

3-18 (a) 2.51 x J0- 8 J ; (b) counterclockwise.
3-19 CcYf"/T.
3-22 2.8 X 104 J.
3-26 (a) 60 J ; (b) 70 J are liberated; (c) Q,., - SO J, Q, .• ~ 10 J.
3-27 (a) t.v•., - Q,.,- 1001, t.u,_, - 900J,
1000 1, - t.u,.• -Q, .• - w._,-
AUc rcle ·- 0 , Q C)'dO - wcr~lc = -500 J.
3- 28 (a) Q - n[a(T2 - T1) + b(Tl - TlJ + c(I/T2 - I/T1)];
(b) f 1, - a + b(T, + T1) - c/T1 T1 ; (c) 24.0 x 10' and 26.0 x 10 3 J kilomole-• K- 1.
3-29 (a) O.S89 J kilomoJe-•K-1; (b) 73.6J kilomoJe- • K-1; (c) 18SO J ; (d) 37.3J kilo-
mole- • K - 1•
3-30 (a) 118 J; (b) 124 J; (c) 118 J .
dJ
3-31 (a) C - 9'dr,
3-32 (b) 1.39 X 104 ] ,
3-33 (a) 1.24 X ICJ3 J ;(b)4000 J ;(c) 1.16 X IO' J.
3-35 (a) -S.JS X 104 J ; (b) W, - -5.25 X 10' J; (C) W4 - -0.98 X ICJ3 J.
3-36 (a) -3.6 X 10' J kg-l; (b) -4.22 X 10' J kg-l.

Chapter 4
4- 2 (a) a.
4-3 (b) S/(3(T, + T,)).
4-4 (a) a = 24.0 J kilomole-1 K-1, b - 6.9 x JQ-3 J kilomole-1 K- 2 ; (b) 2.03 x 104 J
kilomole-1• ·

4-1 (a) 27 x 10' : 4.02 x JQ- 1 ; (b) ~ R:R ; (c) 0.60 ; (d) almost all.
4-8 (b) a+ R.
4-ll (a) q,_,_, - 19 RT1 /2, q,_
4 _, - 17 RT1 /2, q,_, - 9 RT,; (b) 3 R.

4- 16 !!. T - (2n,.nu - n~ - n1Jafc, V(n,. + nJJ)2.

4- 18 (a) a/(c,rr); (b) c,T- 2afv- RTv/(v- b);
(c) (2av(v -b)' - RTv'bJfcp(RTv' - 2a(v - b) 2) .
4-21 (a) nc.Tof2; (b) 3Tof2; (c) 5.25 T0 ; (d) 4.75 nc.T0 •
4-22 885 K.
4- 23 (a) Wr " - 3.46 x 10' J, Ws - -2.5 X 10' J ; (b) Wr - -3.46 X 103 J ; Ws-
-4.43 X 10' J.

Process V,(m 3) T,(K) W(J) Q(J) W(J)

a 32 400 6.73 X 10 1 6.73 X 101 0

4-24
b 13.9 174 2.74 X 101 0 -2.74 X 101

c 32 400 0 0 0
 438 ANSWERS TO PROBLEMS

<4-lS (b)

Process 4T(K) 4Y(m') 4P(atm) W(J) Q(J)

T- const 0 22.4 -O.S J.S7 X 10' J.S7 X JO'

P - const 273 44.8 0 2.27 X 10' S.68 X 101

Y -cons! -438 0 -MOl 0 -S.4S x 10 1

Q -o l6S -67.2 0.901 - 2.04 X 10 1 0

Cycle 0 0 0 1.8 X 101 1.8 )( 10°

Process 4U(J) 4/f(J)

T• const 0 0

P • const 3.41 )( 10' S.68 X 10'

V • const -S.4S X 10° - 9.09 X 10°

Q -o 2.04 x 10° 3.41 )( 10°

Cycle 0 0

<4-26 (a) T(v - b)'.IR -constant, (P + afo'Xv - b)'•.+RIIR - constant,

(b) W • c.(T, - T1) + (afv, - afo1).
<4-30 (a) 900 Calories ; (b) 1600 Calories; (c) 300 and 400 Calories.
<4-31 (b) lower T1 •
<4-32 '1C • T,fT 1•
<4-33 73 K, l30 K.
<4-34 (a) 0.2S, 3; (b) 0.167, s.
<4-36 (a) 2.34 x 105 watts; (b) S.S ; (c) 1.52 x 10' J ; (d) 6.06 x 107 J.
4-37 13.6
<4-38 3. 1 watts, about 0.3Y..

Chepter 6
5-l 83.3 K and 166.6 K .
5-3 (a) 12.2 1 K-1 ; (b) 6.06 x 10' J K- 1•
5-4 (a) Q,., • 2192J, Q... •I0,966J, Q,_. • -6S16 J, Q4 .• • -S480J ;
(b) 0.996 x IO'Nm-•; (c) s•.,
•S.S4 JK- 1, s...
= II.OJK-1, S,.4 • -S.S4JK- 1 ,
S, .• • - II.OJ K- 1•
5-5 (a) 0; (b) 0.167 J K- 1•
 ANSWERS TO PROBU:MS 439

H 293JK-•.
s-7 (a) 1200 J absorbed at 300 K, 200 J given up at 200 K; (b) -3 J K-1, -I J K-1 ,
4 J K- 1 ; (c) 0.
s-s (a) m J K-•; (b) -1n J K- •.
H (a) 0.171 J K - 1; (b) -0.1 71 J K - 1•
5-10 (a) am In (T,}T1) + bm(T, - T1); (b) 2.47 x 101 J kilomolc-• K - •.
s-13 (a) engine; (b) 250R J, -IOOR J ; (c) 0.6; (d) 0.667.
s-ts
As••0 ,(J K-1) AS...,(J K - 1) AS.(J K-1)

(a) 6.93 -5.0 1.93

(b) 11.0 -6.67 4.33

(c) -6.93 20.0 13.1

s-16 (a) (J,$11, 0 - tlOOJ K- AS..,- - 1120 1 K- 1, AS.- 1801 K- 1 ; (b) (),Su,o
1, a
1300 JK- 1,AS,., • - 1210JK-I,S• a90 JK-I.
5- 17 290 K, 190 J K- 1•
s-20 (c) 7i or part (b).
s-22 -0.533 RT, ~ w, S 0, 0 S Au S 0.33:5 RT1 , 0 S AS S 0.693 R.
s-21 No.

6-1 (a) P•v - T{Jv; (c) 0.

6-2 (a) 3360 J kilomolc-• K-•; (b) 0.133.
6-3 (a) R; (b) R In vfv0•
. I
6-4 (a)(),$- 3 aVT + rf P dV +constant; (c) .(T).

6-7 (a) - (T{i - 1)/•; (c) 0.

6-17 3.68 J K-1•
6-18 (b) (c. + RXT- T1) + h,, c1.(T- T0) + ho-
6-19 (a) a + bT- R; (b) 1 - a In <nTo) + b(T- T0 ) - R In P/P0 + 1 0,
h - a(T- T0 ) + b(T' - 71>/2 + ho; (c)(a - R)(T- To) + b(T' - 71)/2 + uo.
6-20 (a) 3.73 x to• J ; (b) us x tO' J K - •.
6-22 (a) -4.6 Jkg-1 ; (b) - I 53 Jkg- 1; (d) 0.394 K .
6-25 (a),..... -0.22 K; (b) 0 ; (c)....., 3.3 K .
6-26 (a) -253 J ; (b) -253 J ; (c) -91 1.
6-28 (a)~ • 0, I' ~ -b/c1. ; (b) 'I • -b/<,.1' = 0.
6-29 -2.1 K.
 440 A NSWERS TO PROBLEMS

6-30 (a)&T - O,Il>s • l.91 x 104 J kilomole- • K- 1 ;(b)ll.T• - 146 K,lu - 6.1 x 10'
J kilomole-• K- 1•
6-32 19.3 atm.
6-33 (a) 0.02 K atm- 1 ; (b) 0.098 K atm-•; (c) - 0.27 K, 12.3 K.
6-34 3S.3 K.
6-38 (oV{oM)s,T • M V/C~.nR
Ch• pter 7
7- 6 (a) P(v +A ) • RT, s • - R In (P/P0 ) + A'P; (b) h • P(A'T- A),
u - T(A'P - R), f- Rn ln (P/P0 ) - I ); (c) cp • PA"T, c, - 2A'P + A• r p-
P'A'' RT
R - R ; (d) • - P(RT _ AP). P- (R- A 'P)/(RT - AP) ; {e) I• - (A - A'T)/PA•r.
7-ll (b) -IO' J, -SO J K- 1, - I.S X IO' J, - 1.48 X IO' J , -800J,3.6 J K-•.
7-23 (a) -J.JS x IO'Nm-• K- 1 ;(b)268atm ; (c)J.31 x IO'Nm- • K- 1 ;{d)24.6atm.
7- 2S (a) 200 K, 1.01 atm ; (b) 1,. - 0.492 J kilomole- •, 1,. • 0.328 J kilomole- •, / 11 -
0.164 J kilomole-•.
7- 17 (a) - 0.1 S K .
Chapte r I
8- l (a) (n,. + nJJ)R In 2.
8-2 (a) t• t• j; (b) t• ~.I a tm; (c) - l.S + 10 7 J; (d) +S x 10' J K- 1•
8- S (a) 2; P a nd T.
8~ (c) K is not a function of P and K - ,-•C'!RT,
8-7 2.
8-8 (a) A - T, % Cd; 8 - T; C - T, % Cd; D - 0; E - T, % Cd; (c) k = 22,000 K
kg kilomole-•.
8-10 (a) 1.28 x Jo-2 To rr; (b) 76.3 atm.
8- ll (a) 0.146 J m-•; (b) A • O.OSS J m-•, c_, - 6.82 x 10-• J m-• K- 1, s - ~.28 x
J0--4 J m-• K-1 ; (cl2.S K.
8-13 (a) du(A 2 - A 1); (b) l(A 1 - A 1) .

8- IS (a) c, - c1 •I•'Tf•, • • -
1
(!!.) ,•• -(.!!...) ;
1 ar,
1
1 a~ r
(b) r,/r1 • •/•, .

8- 17 (a) 4.2 x l o-1 J K-1;(b) 12.6 J ;(c)20.3 J ;(d) -7.7J.

8-18 (c) &G- - 20.3 J, ll.H • - 7.74 J.
8-19 - 228 X 10' J.
du 4
8- 20 (b)) (V1 - V1); (c) l u( V1 - V1).

8- 22 (a) 378 K ; (b) 2.04 x 10-• and S.l4 x 10--4 N m- l,

8-24 (a) bff/T.
8- 2S (a) - 0.8JS J ; ( b) - 1.631, -1.63 J,O, -0.81S J ;(c) 1000c;(d)7.93 x 101•
 ANSWERS TO PROBLEMS 441

8-28 (g) 8.00 X 10' J; (h) -2.02 X 10' J, - 3.96; (i) 5.98 x 10' J ; (j) 0.300; (k) 5.48 X
10' J, 0.275.
8-29 (a) 220 Btu lbm- 1 ; (b) 70 Btu Ibm-•.
8-30 (c) c - 8.7.
Chapter 9
g_1 (a) 3.2 x 10" molecules; (b) 3.2 x 1010 molecules.
g_2 3300 A.
9-3 (a) 6.9 x JQ-•; (b) same as (a).
9-4 (a) 1.7 x JO->; (b) 2.8 x Jo-•.
9-5 (a) O.oJ; (b) 1.7 x JQ- 7 , 2.8 x JQ-•; (c) 1.64 x 1018 v0 molecules m-• s-•, 9.4 X
1020 molecules m-2 s-1.
9-6 (a) 20ms-•, 20ms-1; (b) 12.5ms-1, 14.6ms- 1;,jcl !Oow- 1, 12.2ms-1 ; (d)
10m s-•, 14.1 m s-1 ; (e) li.S m ,-•, 12.7 m s-1.
9-7 (c) 2 vof3 ; (d) 0.707 v0 •
9-8 (b) 3.4 x IO" v0 molecules m- • s-1 ; (c) 4.5 x IO" c0 molecules m-• s-1•
9- 10 Force per unit length - n'mvt/ 2.
9-ll (a) 1360 m s-1 ; (b) 2400 K; (c) 0.31 eV.
g_12 (a). 2.9 x IO" impacts s-1 ; (b) 120m.
g_13 (a) 7.2; (b) 1.22 x to-• atm.
9- 14 (a) 2, x 10" molecules em-•; (b) 3.3 x 1023 impacts s-1 ; (c) same as (b); (d) mean
energy is about 0.1 of heat of vaporization per molecule.
9- 15 (a) 9.4 x JQ-• g em- • , - •; (b) about the same.
g_J6 2.77 VfVA .
9- 17 (a) 1017 molecules; (b) 1.6 x to-3 Torr.
1',
9-18 1'1 = l [I + cxp ( -i!At/2 V)].
9- 20 (b') v,m, «: 1' 11' .
9-21 (a) 3 translational, 3 rotational, and 2(3N - 6) vibrational; (b) 9R, I. II.
9- 23 (a) 1.5 x 10' J ; (b) 1.36 x 10' m s- 1•
Chapter 10
lG-3 (a) 3.2 x w-" m'; (b) 5.2 x 10-• m; (c) 9.6 x 10' s-1•
lG-4 I oc p-t, z oc /'.
JG-5 (a) 5 x 10-10 m; (b) 7.9 x Jo-11 m'; (c) 7.9 x 10' m- 1 ; (d) 7.9 x 10' m- 1 ; (c)
0.45; (f) 0.88 X 10- • m ; (g) 1.3 X Jo-1 m.
JG-6 4.4 X 10- s S.
I G-7 (a) 3.2; (b) 0.05.
IG-8 (a) 3.7 X 10'; (b) 1.35 X JO'; (C) 1.8 X 10'; (d) 1.8 X 10'; (e) 7.4 X 10';
(f) l.S X 10'; (g) - 0.
 442 ANSWERS TO PROBLEMS

l!HI (a) 6; (b) 6; (c) 6.

Io-10 1.2 x to-10 m.
1o-11 (a) 10 em; (b) 6l 1•A.
Io-12 (a) 4.9 x to-10 m; (b) 160; (c) 2.7 m s- 1; (d) 160; (e) 48.
Io-13 (a) 7.2 x to- 14 s; (b) 7.78 x to-• m, 34 atomic distances; (c) 0.2; (d) 632 s.
lo-IS 4 X 1~.

to-16 (a) •li.Jr.,. - 9.6 x to-• N s m'"' K- 111; (b) 4.2 x w-•• m; (c) 2.8 x to-•• m,
2.1 x 1o-10 m.
to-17 (a) A oc T' 11 ; (b) 0.058 J K- 1 m-•
s-•.
Jo-ts (a) 2.52 x to-< m' . - •, 1.03 x 1~ m' s- 1; (b) D oc T' 12m' 11 1• r
Io-1!1 (a) -1.22 x 1015 (moleculesm-•)m-1;(b)(2.32 x IOU +4.75 x 10 11)molecules
s-•; (c) (2.32 x IOU - 4.75 x 1011) molecules s-1; (d) 9.50 x 1016 molecules s- 1,
0.70pgs-1•
to-zo (a) 1.26 x w-•
N s m-•; (b) 0.98 x 1~ m' s-1; (c) 9. 1 x to-• J m- • s-1 K- •.
Chapter 11
ll-3 10'.
11-4 (b) 45, so, 120, 7$, 60, 100.
11-5 (a) S; (b) S, 4, 3, 2, I; (c) 16, 32, 24, 8, 4; (d) IS, 84.
11-6 (a) 6.SS X 10'; (b) 1.52 X 10'2; (c) 2,.
11- 8 ( b) 2427; (c) 3.68, 1.79, 0.838, 0.394, 0.189, O.o78, O.oJS; (d) 7.00.
11-9 (a) ~ 4 macrostates; (d) 2.584, l.S8S, 0.877, 0.485, 0.250, 0.135, 0.058, 0.027.
11-10 (b) 6; (c) 36.
11-12 (a) 8 macrostates ; (d) 2.278, 1.722, 1.056, 0.667, 0.222, 0.056.
11-13 (d) 2.500, I.S91, 0.95$, 0.530, 0.265, 0.114, 0.0378, 0.0075.
11-14 (a) 6N1 - 6N1 - I ; (b) S.SS x 1011, 4.13 x 1012
11-15 (a) 729; (b) 60; (c) 6<; (d) 126.

Macrostates I 2 3 4 s 6

B-E 4S so 120 7S 60 100

u-t 6 rr. F·D 0 0 60 0 12 6
M-B 4,500 2,400 10,800 400 3,840 4,320

j 4 3 2 I

B-E 0.744 1.333 2.100 0.822

F-D 0.769 1.38$ 1.923 0.923
M-B 0.861 1.362 1.694 1.083

ll-17 0.423 k 8 , 0.797 k 8 , 0.539 k 8 •

11-18 (b) 0.395 k 8 •
 ANSWERS TO PROBLEMS I 443

11-19 (b) 3; (c) 19S; (d) 2.923, l.38S, 0.462, 0.231; (r) -2.06 k 8 •
11-21 (b) 12 ; (c) 2.7S, !.SO, 0.1S; (e) -1.81 k 8 •
11-21 (b) 8SOS; (c) 2.86, 1.43, o.m. 0.143; (d) - 3.4 k 8 •
11-29 (a) 4 x 1o-•ev, 6.Sl k 8 ; (b) 127 K; (c) 2 + exp ( -23.2/T); (d) 4 x 1o-•ev,
4.43 k 8 , 14.4 K, I + 2 cxp ( -23.2/ T).
11-34 (a) I + exp ( -</k8 T); (b) [I + exp ( -</koT)j1, [I + exp (</koTJI';
(c) N<[l + exp (</k 8 T))-1 ; (d) Nk8 In [I + exp( -c/k 8 T)) + Nc(711 + exp (</koTJ))- 1 ;
(e) Nko(</ks T)' exp (•/k 8 T)[I + exp (<fk »T)I'.
11-35 (a) E- -pJII'0N/6, M = Np/3; (b) tJ.E - ,..JII'0N/12, tJ.M - 0 ; (c) tJ.M -pN/3.
11-36 (b) U = 0, S - N(pJII'f2T) tanh (!'JII'fk 8 T) + Nk 8 In 2 cosh (pJII'fk8 T). F• •
-Nk 8 Tin 2 cosh (p.JII'fk 0 T), M • - N(pf2) tanh (p.JII'f2k0 T); (c) Ntanh (pJII'f2ksT>.
N [l -tanh (pJII'f2k8 T)).

::~:··~ : Nk[ln V + !In T +!In 2";k + ~J.

1

12-2 (b) ' • Nk7]A.

Alwmk!"]
12-3 (a) Cv • Nk; (b) S - Nk [ 2 + In -;:n,tJ
12-4 (a) 1.2S x 10" molecules; (b) 2.6 x 10" molecules, (c) S.4 x 1011 molecules;
(d) 2.0 x 10" molecules.
12-5 (a) 0.83 v,.; (b) 0.83 v.,.
12-1 (a) 2.08 X IQ-3; (b) 8.3 X 10->; (c) 9 X 1o-•.
12-8 (a) v., • 394m s-1 ; ii • 44S m s-•, v,... • 482 m s-•; (b) 227m s-1, 719 m s-•,
2270 m ,-•.
12-11 (b) <,. - kT/2, i • 3kT/2.
12-12 (c) 0.421 ; (d) 0.079; (e) 0.500; (r) 0.843.
12-13 (c) O.S73; (d) 0.427; (c) 1.00.
12- 16 3.6 x 1o-• m.
12-17 3.26 s.
12-18 (a) 198m ,-• ; (b) 13.S mg hr1 ; (c) 118 s.
12- 19 (a) S.81pgs-1 ; (b) 3.49 x 10 11 molecules s-•; l.l71•1•gs-•; (c) 1.36 x 10' mole-
cules; (d) 3.26 x 10- • Torr
12- 20 0.086 mm, 2.S x 1o-• dcg.
12-21 (a) 6.34 x 1013 neu trons m ·•; (b) 2.63 x to-' N m-•.
12-27 (a) lo-"'; (b) J.S7 x 10' K.
12-29 kT.
12-30 (a) 3kT.
12-31 (a) 12; (b) 9kT, 1.11.
12- 36 (a) 86S, 117, 16; (b) 149 kOv lb•
 444 ANSWERS TO PROBLEMS

Choptor 13
13-2 (a) 246 K ; (b) 172 J kilomole- 1 K-1, 24.9 x 10' J kilomole- 1 K- 1.
13-3 1.12 x 10' J kilomole-• K- 1, 2.66 x 1o-• J kilomole- 1 K- 1.
13-4 Cy - (6Nk1/h•~)T, Cy - 3Nk.
13-5 (c) 2.24 x 10' m r 1; (d) 292 K, 6.1 x 10'' Hz; (e) 3.69 x lo-10 m , 2.27 x
l o-l•m.
1~ (8)6.17 X lo-<'J s4 m-1,4.8 X l o- 11 K s;(b) 7.62 X l o-11 J m-• K-".
13-7 (c) I a-' m.
13-11 (a) 2.24 x 1011 atoms; (b) 1.66 x 10'0 atoms; (c) 2.08 x JO-• oecm', 1.54
Oe cm0•
13-13 S - NJSuJf' tanh (p8 .;t'fkT)/T- Nk In 2 cosh (p 0 .;t'fkT). Cy - Nk(J•uJf'/kT)'
tanh (piJ.;t'fkT).
13-14 0 .1S vF, 0.71 v,..
l .S vj:1.
13-16 18.7 X Jo-lt J ; (c) 1.09 X lo-1 R.
13-17 (a!_l.4 X 10' ~ s-1, 1.3 X lo-t' kg m s-1, 6.S X 10' K; (b) 8.9 X 10-4,6.4 X JQ-4,
2.1 X 10 , 8.9 X 10 , 8.9 X Ia-'; (c) 3200 K .
13-20 (c) •,/3.
13-21 (a) 2.13 x l a-' eV; (b) 2.46 K, 11 6 m s-1.
13-22 2.81 x 1o-u m' N- 1; 3.4 x Jo- 1 K- 1.
 Index

Absolute temperature, 13, 124, 166 Carnol cycle, Il l

Absolute zero, 127 with ideal gas, 112
en1ropy at, 196, 325 with phase lransition, 235
third Jaw, 196 with radiant energy, 244
unauainability or, 199 with surrace film, 243
Adiabatic, boundary, 7, 75 temperature-enlropy diagram, 132
compressibili ty, 157 and thermodynamic temperature, 124,
demagnetization, 231, 404 146
e•pansion, 325 ror three-variable system, 171
first law, 72 Carnotengine, 113, 140, 145
inaccessibility, 172 Carnot rerrige rator, II 5
isentropic, 130 Celsius temperature, 13
processes, 72, I 00, I08, 130, 23 1 Ctntigrade temperature, 13
work, 72, 110, 262 Characteristic equation, 183, 337
Answers to problems, 435 Characteri5tic temperature, 373
Arithmetic mean speed, 2S6 Debye, 393
Assembly, 307 Einstein, 386
Atmospheres, law or, 369 ror linear oscillator, 373
Average speed, 256 ror rotation, 377, 378
Avogadro's number, 370 ror vibration, 377, 378
Characteristic variables, 182
Barometric equation, 369 Charge flow, 41, 67, 223, 285
Bernoulli's equation, 89 Chemical equilibrium, 16, 196, 214
Blackbody radiation, 225, 395 Chemical potential, 207, 209, 214
Bohr magneton, 399 and statistics, 330, 336
Boiling, 38, 243 Classical, dislribution ru nction, 333
Boltzmann constant, 261, 339 C lassical statistics, 345, 350
Boltzmann statistics, 320 Classical theory o r specific heat capacity,
Bose·Einstein s1a1istics, 312 or gases. 267
Bose-Einstein distribution runction, 327 or solids, 271
applied to phonons, 39 1 Clausius-CI•peyron equation, 193 I
applied 10 photons, 395 Clausius inequality, 144
Boundary, 3 Clausius statement or second law, 138
adiabatic, 7, 1S Closed sy5tem, 3, 137
diathermal, 7 Coefficient or expansion, 4S, 46 (su
Boyle's law, 27, 167 Expansion coefficient)
Bridgman me thod, 419 or perrormance, II 5
Brillouin runction, 415 or selr-diffusion, 294
British thermal unit, 78 or thermal conductivity, 292
or viscosity, 266
Calorie, 78 Collision cross section, 279
Calorimetry, 81 macroscopic, 280
Caratheodory principle, 172 microscopic, 280

445
 44& INDEY

Collision frequency, 284 Derivatives, partial, 42

Collisions, wrth moving wall, 262 second-order partial, S3
with wall, 2S4 Dewar flask, 8
Compressibility, adiabatic, l S7 Diather mal boundary, 7
of copper, 46 D iatom ic gas, S1, 269, 376
of electron gas, 416 Dielectric work, 6S
of helium, IS I D ieterici equation of state, 60
of ideal gas, 47 D ifference between specific heat capacities,
isothermal, 46, 47 99, lSI
of mercury, 46 D ifferent ials, exact, S3
of van der Waals gas, S9 inexact, 69, 77
Condition equation, 2 13, 421 D iffusion, 294
Conductivity, electrical, 28S D isorder. 324
thermal, 292 Dissipative work, 71, 78
Configuration work, 70 D istinguishable paNicles, 308, 320, 334,
Conservation of energy, 87 39 1,399
Constant volume gas thermomecer, 9 Dislribut ion function, Bose-Ejnstein, 327
Constituents, 210 classical, 333
Conversion factors, back cover Fermi-Dirac, 33 1
Cooling, magnetic, 231 Gaussian. 360
Copper, f(>mpressibility, 46 Maxwell- Boltzmann, 334, 430
Debye ~emperature, 393 speed, 3SS, 362
specific heat capacity, 82, 16-. velocity, 3S9
thermal expansion, 46 Drude theory, 28S
Corruponding statu, Sl DuLong-Petit Law, 81,271,394
C ritical constants, table, 36
van der Waals gas, 49
Cri tical pressure, 3S Efficiency, 113, 140, 145, 239
Critical point, 36 Einstei n theory of specific heat capacity,
Critical temperature, 3S 386
Critical volume, 3S, 278 Elastic waves, 388
C ross section, 291 Electrolytic cell, 223
Curie constant, 401 equatton of state, 4 1
C urie's law, 4 1, 230 work, 67
statistical derivation, 401 Elcetron gas, distribution function, 331
cycle, Carnot, Ill, 23S thermodynamic prope rt ies, 407
Rank ine, 239 Elect ron-volt, 262
Cyclic process, S4, 69, n Electronic conduction, 285
Electronic mea n free path, 283
Dalton's law of partial pressures, S7, 264 E mpirical temperaiU re, 9, 124, 166
Danie ll cell, 223 Energy, free, 179, 180
Debye temperature, 393 internal, 73, 98, 270
Debye theory of specific heat capaclly, 387 kinetic, 86, 261 , 266, 270
Degeneracy of a macrolevel, 3Y. levels, 305
of a state, 306 potential, 86, 184, 228, 266, 366, 403
Degrees of freedom, 266, 370 states, 305
rotational, 266 total, 86, 228, 403
translational, 266 Energy equation, 98
vibrational, 266 for ideal gas, lOS
Demagnetization, adiabatic, 231 , 404 surface, 98, lOS
Density, 4 for van der Waals gas, I I S
reduced, I SO Energy, equipartition of, 264, 370
 INDEX 447

Engine, Carnot, 113 surface, 41

steam, 233 Equations of stare of gases, D ieterici, 60
Enginuring applications, 233 Hirn, 277
Ensemble, 307 ideal gas, 25, 251,-258, 353
average, 311 kinetic theory, 251,258,277
Enthalpy, 84, 100 statistical thermodynamics, 353
characteristic equation, 183 van der Waals, 28, 277
and heat flow, 100 virial, 30
heats of transformation, 84 Equilibriu m, chemical, 16
Joule experiment, 107 mechanical, 16
statistical expression, 340 metastable, 186
thermodynamic potential, 181 stable, 124, 186, 427
th ird law, 196 thermal, 16
Enthalpy of ideal gas, 159 thermodynamic, 17
pure substance, 157 unstable, 186
solid, 174 Equipartition of energy, 264, 370
water, 100, 234, 235 Error function , 380
van der Waals gas, 174 Escaping tendency, 214
Entropy. 127 Escermann, Simpson, and Stern, cxperi·
at absolute zero, 198 men!, 242
disorder, 324 Exact different ial, 53, 170
increase of, 123, 135 Exclusion principle, 317, 408
irreversibility, 135 Expansion coefficient, 45
Nemst heat theorem, 198 of copper, 46
principle of increase of, 123, 135 of electron gas, 416
reve rsibility, 136 of he lium, 150
second law, 123 of ideal gas, 45.
statistical interpretation, 323, 427 linear, 46
thermodynamic stability, 178, 188 of mercury, 46
third law, 196 of solid, 173
En tropy diagrams, h-s-P, 234 of va n der Waals gas, 59
Mollier, 235 volume, 45
T-S, 132 Expansion, free, 70, 325
Entropy of Einstein oscillators, 413 Extensive variable, 3
electron gas, 412
ideal gas, 159, 354
mixing, 208 Factorials, 424
mullivariable syste ms, 170 Fahrenheit temperature, 14
open systems, 209 Fermi energy, 408
paramagnet, 229 Fermi temperature, 415
pure substances, 157 Fermi velocity, 415
solids, 163 Fermi-Dirac distribution function, 33 1
van der Waals gas, 161 408
Equations of state, 24 Fermi-Dirac statistics, 317
Curie's law, 41
First Jaw of thermodynamics, 73
oieleclric, 41
electrolytic cell, 42 analytic form, 76
combined with second law, 148
liquid, 48, 163
mullivariable systems, 184 general form, 86
paramagnet, 41,401 First-order phase transitions, 192
radian t energy, 227 F ixed points, 15
solid, 48, 163 table, 16
 448 INDEX

Flow processes, Joule-Thomson experi- inexact differenlial, 77, 149

ment, lOS irreversible, 133
nozzle, 89 reversible, 131
steady flow, 87 statistical interpretation, 326
turbine, 89 and third law, 196
Free energy, 179, 180 Heat reservoir, 83
Free expansion, 70, 104, 325 Heat of transformation, 83, 19 1
F reezing, 38, 84, 188, 194 fusion, 84 ·
fundamental constanls, back cover lambda, 192
Fusion, heat or, 84 sublimation, 84
or a surface fi lm, 219
Gamma function, 424 vaporization, 84, 192
Gas constant, universal, 2S, 260 or water, 8S, 130
Gases, electron, 407 Helium, critical constants, 36
ideal, 25, I59, 258 density, lSI
photon, 395 lambda transition, 33, 192
van der Waals, 28, 160, 276 P-o-T surface for, 39
vapor, 36 specific heal capacity of, l SI
Gaussian distribution, 360 thermodynamic properties or, 150
Gay-Lussac- J oule experiment, I 03 three, 194, 202, 416
Generalized Helmhollz function, 184, 229, van der Waals constants or, '28
341, 368, 401 Helmholtz function, 178
Gibbs-Duhem equation, 241 characteristic variables. 183
Gibbs function , 180 generalized, 184, 341
characteristic variables, 183 open systems, 210
chemical potential, 206 and partition function, 337
open systems, 210 stable equilibrium, 179, 188
and partilion function, 339 statistics expressions, 337
phase transitions, J90 Helmholtz function of electron gas, 412
stable equilibrium, 180, 189 ideal gas, 181
statistics expression, 339 paramagnet, 229
third law, 197 _ radia nt energy, 228
Gibbs function of ideal gas, 180, 33' surface film, 220
radiant energy, 228, 398 van der Waals gas, 181
Gibbs- Helm holtz equations, 183 Hill and Lounasmaa data, ISO
G ibbs paradox, 241 1•-s-P surfaces, 234
Gibbs phase rule, no chemical reactions, Hydrostatic pressure, 4, 163
214
chemical reactions, 216
Grand partition function, 347 Ice, phase diagram, 35
Gravitational field, 366 point, 12, 195
Ideal gas, 25
Heat (.-~ Heat flow) latent, 83, 191 adiabatic processes, 108, 262
mechanical equivalent of, 77 and Boflzn1ann constant, 260, 339
or tra nsformation, 83, 191 and Carnol cycle, 112
Heal capacity, 80 (stt Specific heat compressibility, 47
capacity) diatomic, 267, 376
Heat engine, 11 3, 233 energy equation, lOS
Heat flow, 74 enthalpy or, 108, 159
and absolute temperature, 124 entropy or, I59, 3S4
and enthalpy. 100 equation or state, 2S, 258, 3S3
tnd entropy, 128 expansivity, 45
 INDEX 449

Ideal gas (contd.) Inversion point, 107

Gibbs function of, 180 Inversion population, 407
in gravitational field, 366 Inversion temperature, 16S
Helmho ltz function of, 181 Irreversible proce.u, 18
internal energy of, lOS, 261, )SJ dissipative work, 71
isothermal work, 64 entropy changes, 133
J oule coefficient, 104 entropY. production, 136
Joule-Thomson coefficient, I08 and Gobbs function, 180, 189
kinetic theory, 2S1 and Helmholtz function, 179, 188
monatomic, 269 Rank ine cycle, 240
partial pressure, 207, 264 throllling. 106
P-v-T surface for, 26 Isentropic process, 100, 108, ISS
Sackur-Tetrode equation, 3S4 isothermal, 170
specific heat capacity, 267, 3S4, 376 on multivariable systems, 170
statistical thermodynamics, 339, ·JSO on paramagnets, 230
temperature, 113, 121, 167 second law, 130
Increase of entropy, 123, IJS third law, 199
Inequality of Clausius, 144 Isobaric process, 18, 26, 101
Inexact differential, 69, n entropy change in, 131
lnftection point, 49 expansivity, 43
Integrating denominator, 169 phases transitions, 37, 187
Intensive variable, 3 work, 64
' ntermolecular forces, 276, 278 l sochoric process, 18, 26, 99
'nternal energy, 74 entropy change in, 130
chemical potential, 209 work, 64
energy equation, 98 Isolated system, 3, I 36, 188
isolated system, 188 Isothermal compressibility, 47
kinetic theory, 262, 267 Isothermal process, 18, 26
open systems, 209 compressobility, 47
and parrition function, 338 en tropy change, I30
and specific heat capacity, 99 isentropic, 170
statistical interpretation, 326, 338 phase transitions, 188
surface, lOS work, 6S
thermodynamic potential, 181 lsovolumic processes, 18
and total energy, 88, 184
nternal energy of blackbody radiation, Joule coefficient, 104, 164
276, 396 experiment, 104
Debye solid, 386 Joule-Thomson coefficient, I 07, 164, 200
Einstein solid, 386 experi ment, 106
electrolytic cell, 224
•lectron gas, 410 Kelvin absolute temperature, 13
ideal gas, 104, 3S3 Kelvin-Joule experiment, 107, 164
linear oscillators, 37S Kelvin-Planck statement or second law,
paramagnet, 239 139
surface ftlm, 219 Kelvin temperature scale, 13, 126
van der Waals gas, II S, 161 Kilomole, 4
water, 84 Kinetic energy of systems, 86
nternational practical temperat ure scale, or particles, 261, 3SS
IS and thermal conductivity, 292
nternational steam table calorie, 79 Kinetic theory (stt Chapters 9 and 10),
nvarianl systems, 21 S assumptions. 2S I
nversion curve, 107, 16S heat capacity, 267,271
 450 INDEX

Kinetic theory (contd.) Mean square speed, 259

ideal gu, 258 . Mechanical equilibrium, 16
transpon coefficients, 286, 292, 294 Mechanical equivalent of heat, 77
van der Waals gu, 276 Mercury, compressibility, 46
expansivity, 46
Lagrange method of undetermined multi- specific heat capacity, 82
pliers, 421 · van der Waals cons1an1s, 28
Lambda point, 40, 192 Metlllurgical limit, 238
Latent heat, 83, 191 (su Heat of trans- Microscopic property, 2, 302
formation) Microstate, 307
Law of atmospheres, 369 Miller and Kusch experiment, 365
conservation of energy, 87 Mixing, entropy of, 208
corresponding states, 51 Molal specific value, 4
Laws of thdrmodynamics, first, 73 Mole fraction, 206
second, 123, 138, 139, 172 Molecular beam, 362
third, 198, 199 Molecular diameter, 278, 290
zeroth, 6 Molecula r nux, 254, 289
Linear cxpansivily, 46 Molecular weight, 4, 243, 293
Linear oscillator, 372 Mollier diagram, 234
Liquid, 163 Monatomic ideal gas, 267, 350
drop, 221 Monovariant systems, 215
saturated, 34, 188, 238 Most probable speed, 357
Multivariable system, 169, 184, 341
Macroleve l, 350
Macroscopic propeny, 2, 302 Negative temperatures, 405
Macrostate, 307 Nernst heat theorem, 198
most probable, 427 Neutral equilibrium, 192
MagnetiC moment, 399, 432 Nozzle, now through, 89
saturation, 401
Magnetism, equation of state, 41 Occupation number, 307
statistical thermodynamics of, 399 average, 31 I
thermodynamics of, 228 most probable, 430
work, 6S Ohm's law, 286
Magnetic potential energy, 228, 432 Open systems, 3, 206
Maxwell-Boltzmann distribution function, chemical potential, 208
334 phase «Juilibrium, 210
Maxwell-Boltzmann speed distribution and stallstical mechanics, 323, 327
function, JSS, 362 Oscillators, 266, 372, 386
Maxwell-Boltzmann statistics, 320
Maxwell-Boltzmann velocity distribution Paramugnetism, adiabaticdcmagnt:litation,
function, 359 230
Maxwell-Boltzmann distribution function, Brillouin runction, 415
applied to Debye solid, 387 Curie's law, 4 1
applied to Einstein solid, 386 equation or stare, 41
applied to linear oscillator, 372 potential energy, 433
applied to paramagnetism, 399 statistical thermodynamics, 399
Maxwell relations, ISS, 230 thermodynamics, 228
Mean free path, 281 work, 65
electronic, 283 Partial derivatives, 47
table, 290 mixed second-order, 53, ISO
in transport properties, 288 relations between, Sl, 102
Mean free time, 284 Partial pressure, 57, 264
 INDEX 451

Parricles, distinguishable, 307, 320 isobaric, 18, 37, 101, 130

indistinguishable, 307, 312, 317, 333 isochoric, 18, 99, 130
Partition function, 336 isothermal, I8, 170
grand, 347 reversible, 18, 130
ideal gas in gravitational field, 367 quasistatic, 17, 18
linear oscillator, 373 · Property, thermodyna mic, 3, S4
magnetic, 400, 414 Pure subs tance, IS7
monatomic ideal gas, 350 P-IJ-T surface, helium, 39
Path function, 70, 77 idea I gas, 26
Pauli exclusion principle, 3 17, 408 liquid, 42
Performance, coefficient of, I IS solid, 42
Phase diagra m, cadmium-bismuth system, substance contracting upon freeting, 31
242 subs tance expanding upon freezing, 32
helium, 39 van der Waals gas, 29
substance contracting upon freezing, 3 I water, JS
substance expanding upon freezing, 32
water, JS
Pha~ equilibria, Clausius-Clapeyron equa-
Quantum theory, 302
roon, 193 of electron gas, 407
many phases, 210 of linear oscillators, 372
two phases, 190 of pa ramagnetism, 399, 432
Phase ru lei chcmicul reactions, 216 of specific heat capacities, 376, 386, 387
no reac1ions, 21 S Quuistatic process, 17, 18
Phase transitions, 30, I 90
order of, 192 Radiation, black body, 22S, 39S
Phonon gas, 391 thermometry, IS
Photon gas, 39S Rankine cycle, 239
Planck, constant, 304 Rankine tempenuure, 14
Kelvin-statement of second Jaw, 139 Ratio of specific heat capacities, 108
radiati on la w, 226, 396 Rayleigh-Jeans Jaw, 397
third law, 198 Reactions, chemical, 16, 138, 196, 216
Platinum resistance thermometer, 7, 15 Redua:d variables, density, ISO
Population inversion, 407 press ure, SO
Porous pl ug experiment, lOS temperature, SO
Potential ene rgy, 86, 184, 342 volume, SO
gravitational, 366 Refrigerator, Carnot, I l l, 140
mallnetic, 288, 403, 432 magne tic, 230, 404
oscollator, 266 Reservoi r, heat, 83
Pressure, 4 Resistance thermometer, germanium, 8
critical, 3S, 49 platinum, 7, IS
kinetic inte rpretation, 2S9 Reversible engine, II 3, <13
partial, S7, 264 Reversible processes, 18 136, 148
radiation, 226 adiabatic, 108
redua:d, 3S cycles, I I I , 233
statistical interpretation, 326, 338 entropy change in, 130
vapor, 34, 216, 221 hea t now in, 130, 326
Probability, thermodynamic, 310 work in, 63, 72, 326
Process, 17 Root mean square speed 261, 3S8
adiabatic, 18, 72, 130, 231
cyclic, S4, 69, 77 Sackur-Terrode equa tion, 3S4
irreversible, 18, 133, 136 Saturated liquid, 34, 188, 238
isentropic, 130, 170, 230 Saturated vapor, 34, 187
 452 INDEX

Second law of thermodynamics, Cara- Statistical thermodynamics, 2, 2SO, 302,

theodory principle, 172 337
Clausius slatement, 138 Statistics, Bose-Einstein, 312
and first law, 148 classical, 345
increase of entropy, 123 Fermi·Dirac, 317
Kelvin·Pianck statement, J39 Maxwell-Boltzmann, 320
Second·order mixed partial derivatives, 53, Steady now, 87
150 Bernoulli's equation, 89
Second-order phase-transitions, 192 nozzle, 89
Self-diffusion coefficient, 294 porous plug, 106
Shafl work, 87 turbine, 89
Solid, 163 Steam cycle, 235
specific Ileal capacity, 271, 386, 387 Steam point, II
Specific heat capacity, 80 Stefan's law, 227, 398
classical theory, 267 Stefan-Boltzmann law, 227, 398
at constant pressure, 80, 101, JS4 Stirling's approximation, 313, 42S
at constant volume, 80, 99, 152, 161 Stretched wire, 40, 65
of copper, 82 Sublimation, 39, 194
Debye, 387 Sum over states, 336
of diatomic gas, 376 Superconductor, 245
differences, 108, 151 Supercooled, 187
Dulong-Petit law, 81 Superheated, 188
Einstein, 386 steam, 238
of electron gas, 411 Surface film, equation of state, 41
of gases, 108, 267 tension, 218
of helium, I 51 thermodynamics of, 218
of mercury, 82 work, 68
molecular, 293 Surroundings, 3
of a monatomic gas, 267 Survival equation, 281.• 283
o f a paramagnet, 229, 404 System, 3, 307
ratio, 108 closed, 3
of a solid, 271 isolated, 3
of a su rface film , 220 open, 3
of water, 134
Specific value of an exte nsive variable, 3 T ds equations, ISS
Speed, average, 358 Temperature, S
distribu tion function, 3SS, 362 absolute, 13
mean square, 259 absolute zero, 127, 196
most probable, 3S7 Celsius, 13
root mean square, 26 l , 358 centigrade, 13
Sponta neous process, 196 characteristic, 373, 377
Stable equilibrium, 212 critical, 35
and entropy, 188 Debye, 393
and Gibbs function, 189 Einstein, 386
empirical, 9, 166
and Helmholtz function, 188
Fahrenheit, 14
and spontaneous process, 188 Fermi, 415
Standard form for thermodynam ic fixed poin ts, 16
formulas, 149 ice point, 12, 195
State variable, 3 International Practical Scale, 15
extensivej3 Kelvin, 13
intensive, 3 negat ive, 405
 INDEX 453

T emperaiU re (contd.) o f water, I 0, 19S

Rankine, 14 T·s diagrams, 132
reduced, SO T urbine, 89
re ference, IS
steam point, 12 Ultraviolet catastrophe, 397
thermodynamic, 13, 124, 166 U nattainability statement of third law,
triple potnt, 13, 33, 19S 199, 232
T hermal conductivity, 292 Universal gas constant, 2S
Thermal efficiency. 140 per molecu le, 261
Thermal equilibnum, 6
T hermal expansion, 4S Van der W aals gas, constants, 28
Thermocouple, 8 critical constants, 49
Thermodynamic equilibrium, 16 energy equation, 161
Thermodynamic formulas, 4 19 equation of state, 28, 30, 276
Them>odynamic laws, 6, 73, 123, 138, 139, Helmholtz funct ion, 181
172, 198, 199 Joule and Joule·Tho mson coefficients,
T hermodynamic poten tials, 181 16S
The rmodynamic probabili ty, 3 12, liS, 319, kinetic theory, 276
322 law of correspond ing states, Sl
T hermodynamic system, 3 P-v-T surface, 29
Thermodynamic tempent ture, 13, IS, 124, thermodynamic properties, 160
166 Vapor, 36
T hermodynamics, classical, 2 Vapor pressure, 36, 19S
stati<tical, 302 hquid d rop, 221
Thermometer, constant volume gas, 9 pressure dependence, 216
germanium, 8 saturated, 36
platinum, 7, I S supercooled, 187
thermocouple, 8 superheated, 36
Thermometry, Carnot cycle, 124 Vaporization, heat of, 84, 192
constant volume gus, 9 Variable, extensive, 3
ideal gas, II, 127 intensive, 3
radiation, 15 specific, 3
resistance, JO, IS state, 3
thermocouple, 8 , IS Variance, 214
T hermomolecular pressure rat io, 273 invariant systems, 2 1S
Thermoscope, 7 monovariant systems, 21S
Third law of thermodynamics, 196 Velocity, distribution o f molecular, 3S•
and magn et ism, 230, 404 Fermi , 4 1S
Nernst heat theorem, 198 fluid, 87 l
Planck s tateme nt, 198 Maxweii· Boltzman n distribut ion, 3S9
statistical interpretation, 324 space, 3S6
unattainability statement, 199, 232 Viroal coefficients, 30
Transformation, heat of, 83. 19 1 Viscosity, coefficient of, 286
Transport phenomena, diffusion, 294 table , 290, 293
thermal conductivity, 292 Voltaic cell (su Electrolytic cell)
viscosity, 286 Volume, crit ical, lS, 278
T riple point, cell, 10 molecular, 278
and e nthalpy, 8S reduced , SO
and Gibbs function, 191 s pecifoc, 4
heats of tr~nsformation at, 8S
phase equi libria, 21S \Vater, critical constants, 3 ·
table, 33 density, 20, 174
454 INDEX

Water, critical constants (contd.) free expansion, 70

heat of fusion, 141 inexact differential, 69
heat of vaporization, 84, 8S, 130 irreversible, 71, 137
ice point, 12 isentropic, llO
P·IJ-T diagram, 32, 3S isobaric, 64
specific heat capacity, 134 isochoric, 64
steam point, 12 iJothcrmal, 6S
surface tension, 219 magnetic, 6S
triple pqint, 10, 13, l9S maximum, 179
van deriWaals constants, 28 in a phase change, 84
vapor pressure, 218 shaft, 87
Wien's law, 397 statistical interpretation, 326
Wire, equacion of state, 40 surface film, 68
work, 6S van der Waals gas, 163
Work, adiabatic, 72, 110
configuration, 70 Young's modulus, 41
depends on path, 69
dielectric, 67
dissipative, 71, 78 Zartman and Ko experiment, 363
electrolytic cell, 67 Zero, absolute, 127, 196, 199
-energy theorem, 62 Zeroth law of thermodynamics, 6
external, 63 Zustandssumme, 336

.,