Chapter 3

Non-ideal Gases

This chapter is devoted to considering systems where the interactions be-

tween particles can no longer be ignored. We note that in the previous
chapter we did indeed consider, albeit briefly, the effects of interactions in
fermion and in boson gases. This chapter is concerned more with a system-
atic treatment of interatomic interactions. Here the quantum aspect is but
a complication and most of the discussions will thus take place within the
context of a classical description.

3.1 Statistical Mechanics of Interacting Par-

ticles
3.1.1 The partition function
We are now considering gases where the interactions between the particles
cannot be ignored. Our starting point is that everything can be found from
the partition function. We will work, initially, in the classical framework
where the energy function of the system is
X p2 X
i
H (pi , qi ) = + U (qi , qj ) . (3.1.1)
i
2m i<j

Because of the interaction term U (qi , qj ) the partition function can no longer
be factorised into the product of single-particle partition functions. The
many-body partition function is
Z P P
1 − ( i p2i /2m+ i<j U (qi ,qj ))/kT 3N
Z= 3N
e d p d3Nq (3.1.2)
N !h

133
134 CHAPTER 3. NON-IDEAL GASES

where the factor 1/N ! is used to account for the particles being indistinguish-
able.
While the partition function cannot be factorised into the product of
single-particle partition functions, we can factor out the partition function for
the non-interacting case since the energy is a sum of a momentum-dependent
term (kinetic energy) and a coordinate-dependent term (potential energy).
The non-interacting partition function is
Z
VN P
− ( i p2i /2m)/kT 3N
Zid = e d p (3.1.3)
N !h3N
where the V factor comes from the integration over the qi . Thus the inter-
acting partition function is
Z P
1
Z = Zid N e−( i<j U (qi ,qj ))/kT d3N q. (3.1.4)
V
The “correction term” is referred to as the configuration integral. We denote
this by Q Z
1 P
Q= N e−( i<j U (qi ,qj ))/kT d3N q. (3.1.5)
V
(Different authors have different pre-factors such as V or N !, but that is not
important.) The partition function for the interacting system is then
 N
1 V
Z= Q (3.1.6)
N ! Λ3
and the attention now focuses on evaluation/approximation of the configu-
ration integral Q.

3.1.2 Cluster expansion

We need a quantity in terms of which to perform an expansion. To this end
we define
fij = e−U (qi ,qj )/kT − 1, (3.1.7)
which has the property that fij is only appreciable when the particles are
close together. In terms of this parameter the configuration integral is
Z Y
1
Q= N (1 + fij ) d3N qi (3.1.8)
V i<j

where the exponential of the sum has been factored into the product of
exponentials.
3.1. STATISTICAL MECHANICS OF INTERACTING PARTICLES 135

Next we expand the product as:

Y X XX
(1 + fij ) = 1 + fij + fij fkl + . . . (3.1.9)
i<j i<j i<j k<l

The contributions to the second term are significant whenever pairs of parti-
cles are close together. Diagrammatically we may represent the contributions
to the second term as:

Contributions to the third term are significant either, if i, j, k, l are distinct,

when pairs i – j and k – l are simultaneously close together or, if j = k in
the sums, when triples i, j, l are close together. The contributions to the
third term may be represented as:

The contributions to the higher order terms may be represented in a similar

way. The general expansion in this way is called a “cluster expansion” for
obvious reasons.

3.1.3 Low density approximation

In the case of a dilute gas, we need only consider the effect of pairwise
interactions – the first two terms of Eq. (3.1.9). This is because while the
probability of two given particles being simultaneously close is small, the
probability of three atoms being close is vanishingly small. Then we have
Y X
(1 + fij ) ≈ 1 + fij (3.1.10)
i<j i<j

so that, within this approximation,

Z ( X
)
1
Q= N 1+ fij d3N qi
V i<j
Z X (3.1.11)
=1+ fij d3N qi .
i<j
136 CHAPTER 3. NON-IDEAL GASES

There are N (N −1)/2 terms in the sum since we take all pairs without regard
to order. And for large N this may be approximated by N 2 /2. Since the
particles are identical, each integral in the sum will be the same, so that
Z
N2
Q=1+ f12 d3 r12 . (3.1.12)
2V

The V N in the denominator has now become V since the integration over
i, j 6= 1, 2 gives a factor V N −1 in the numerator.
Finally, then, we have the partition function for the interacting gas:
 Z 
N 2  −U (r)/kT  3
Z = Zid 1 + e −1 d r (3.1.13)
2V

and on taking the logarithm, the free energy is the sum of the non-interacting
gas free energy and the new term
 Z 
N 2  −U (r)/kT  3
F = Fid − kT ln 1 + e −1 d r . (3.1.14)
2V

Since U (r) may be assumed to be spherically symmetric, in spherical polars

we can integrate over the angular coordinates:
Z Z
. . . d r → 4π r2 . . . dr
3
(3.1.15)

to give
 Z 
N2 2
 −U (r)/kT 
F = Fid − kT ln 1 + 4π r e − 1 dr . (3.1.16)
2V

In this low density approximation the second term in the logarithm, which
accounts for pairwise interactions, is much less than the first term. — Oth-
erwise the third and higher-order terms would also be important. But if the
second term is small then the logarithm can be expanded. Thus we obtain
Z
N2  
F = Fid − 2πkT r2 e−U (r)/kT − 1 dr. (3.1.17)
V

A more rigorous treatment of the cluster expansion technique, including

the systematic incorporation of the higher-order terms, is given in the article
by Mullin [1].
3.2. THE VIRIAL EXPANSION 137

3.1.4 Equation of state

The pressure is found by differentiating the free energy:

∂F 
p=−
∂V T,N
Z (3.1.18)
N N2  
= kT − kT 2 2π r2 e−U (r)/kT − 1 dr.
V V

We see that the effect of the interaction U (r) can be regarded as modifying
the pressure from the ideal gas value. The net effect can be either attractive
or repulsive; decreasing or increasing the pressure. This will be examined, for
various model interaction potentials U (r). However before that we considered
a systematic way of generalising the gas equation of state.

3.2 The Virial Expansion

3.2.1 Virial coefficients
At low densities we know that the equation of state reduces to the ideal
gas equation. A systematic procedure for generalising the equation of state
would therefore be as a power series in the number density N/V . Thus we
write  2  3
p N N N
= + B2 (T ) + B3 (T ) + .... (3.2.1)
kT V V V
The B factors are called virial coefficients; Bn is the nth virial coefficient.
By inspecting the equation of state derived above, Eq. (3.1.18), we see that
it is equivalent to an expansion up to the second virial coefficient. And the
second virial coefficient is given by
Z ∞
 
B2 (T ) = −2π r2 e−U (r)/kT − 1 dr (3.2.2)
0
which should be “relatively” easy to evaluate once the form of the interpar-
ticle interaction U (r) is known. It is also possible to calculate higher order
virial coefficients, but it becomes more difficult.
138 CHAPTER 3. NON-IDEAL GASES

3.2.2 Hard core potential

(The reader is referred to Reichl [2] for further details some of the models in
the following sections.)
The hard core potential is specified by

U (r) = ∞ r<σ
(3.2.3)
=0 r > σ.

Here the single parameter σ is the hard core diameter: the closest distance
between the centres of two particles. This is modelling the particles as im-
penetrable spheres. There is no interaction when the particles are separated
greater than σ and they are prevented, by the interaction, from getting any
closer than σ. It should, however, be noted that this model interaction is
un-physical since it only considers the repulsive part; there is no attraction
at any separation.
The gas of hard sphere particles is considered in some detail in Section 3.6.
For the present we are concerned solely with the second virial coefficient.

Figure 3.1: Hard core potential

For this potential we have

e−U (r)/kT = 0 r<σ

(3.2.4)
= 1 r>σ

so that the expression for B 2 (T ) is

Z σ
B2 (T ) = 2π r2 dr
0 (3.2.5)
2
= πσ 3 .
3
3.2. THE VIRIAL EXPANSION 139

In this case we see that the second virial coefficient is independent of tem-
perature, and it is always positive. The (low density) equation of state is
then  
2 3N
pV = N kT 1 + πσ (3.2.6)
3 V
which indicates that the effect of the hard core is to increase the pV product
over the ideal gas value.
It is instructive to rearrange this equation of state. We write it as
 −1
2 3N
pV 1 + πσ = N kT (3.2.7)
3 V

and we note that the “correction” term 23 πσ 3 N /V is small within the validity
of the derivation; it is essentially the hard core volume of a particle divided
by the total volume per particle. So performing a binomial expansion we find
to the same leading power of density
 
2 3N
pV 1 − πσ = N kT (3.2.8)
3 V
or  
2
p V − N πσ 3 = N kT. (3.2.9)
3
In this form we see that the effect of the hard core can be interpreted as
simply reducing the available volume of the system.

Excluded volume
By how much is the volume reduced? Two spheres of diameter σ cannot
approach each other closer than this distance. As indicated in Fig. 3.2 this
means that the effect of one particle is to exclude a sphere of radius σ from
the other particle.

Figure 3.2: Volume of space excluded to particle 2 by particle 1

140 CHAPTER 3. NON-IDEAL GASES

Two particles exclude a volume 43 πσ 3 . Thus the “excluded volume” per

particle is one half of this. And the excluded volume of N particles is then

2
Vex = N πσ 3 , (3.2.10)
3
that is, four times the volume of the particles. This is precisely the volume
reduction of Eq. (3.2.9).
We should note that the second virial coefficient for the hard sphere gas
is then simply the excluded volume per particle.

3.2.3 Square-well potential

The square-well potential is somewhat more realistic than the simple hard
core potential; it includes a region of attraction as well as the repulsive hard
core. The potential is specified by

U (r) = ∞ r<σ
= −ε σ < r < Rσ (3.2.11)
= 0 Rσ < r

so we see that it depends on three parameters: σ, ε and the dimensionless R.

Figure 3.3: Square well potential

For this potential we have

e−U (r)/kT = 0 r<σ

= eε/kT σ < r < Rσ (3.2.12)
= 1 Rσ < r
3.2. THE VIRIAL EXPANSION 141

so that the expression for B2 (T ) is

 
 Zσ ZRσ 
2 ε/kT

2
B2 (T ) = −2π (−1) r dr + e −1 r dr
  (3.2.13)
0 σ
2 n

o
= πσ 3 1 − R3 − 1 eε/kT − 1
3
or
n o
B2 (T ) = 2
3
πσ 3 3 3
R − (R − 1)e ε/kT
(3.2.14)
= 23 πσ 3 R3 − 23 πσ 3 (R3 − 1)eε/kT .

In this case, using the more realistic potential, we see that the second
virial coefficient depends on temperature, varying as

B2 (T ) = A − Beε/kT . (3.2.15)

The second virial coefficient for nitrogen is shown in Fig. 3.4. The square
well curve of Eq. (3.2.14) has been fitted through the data with ε/k = 88.3 K,
σ = 3.27 Å (1 Å = 10−10 m), and R = 1.62. Observe that this crude approx-
imation to the inter-particle interaction gives a remarkably good agreement
with the experimental data. The figure also shows the hard sphere asymptote
B2hs = 44.28 cm3 /mol.
At low temperatures, where B2 (T ) is negative, this indicates that the
attractive part of the potential is dominant and the pressure is reduced com-
pared with the ideal gas case. And at higher temperatures, where it is intu-
itive that the small attractive part of the potential will have negligible effect,
B2 (T ) will be positive and the pressure will be increased, as in the hard
sphere case. The temperature at which B2 (T ) goes through zero is called the
Boyle temperature, denoted by TB . For the square well potential

−ε/k
TB =
(3.2.16)
ln 1 − R13

At very high temperatures we see from the expression for B2 (T ) that it

saturates at the hard core excluded volume.

3.2.4 Lennard-Jones potential

The Lennard-Jones potential is a very realistic representation of the inter-
atomic interaction. It comprises an attractive 1/r6 term with a repulsive
142 CHAPTER 3. NON-IDEAL GASES

Figure 3.4: Second virial coefficient of nitrogen as a function of temperature

with the square well functional form Eq. (3.2.14). Square well parameters
ε/k = 88.3 K, σ = 3.27 Å, and R = 1.62.

1/r 12 term. The form of the attractive part is well-justified as a description of

the attraction arising from fluctuating electric dipole moments. The repulsive
term is simply a power law approximation to the effect of the overlap of the
atoms’ external electron clouds. We write the Lennard-Jones potential as
  
σ 12  σ 6
U (r) = 4ε − ; (3.2.17)
r r

this depends on the two parameters: ε and σ as shown in Fig. 3.5.

Figure 3.5: Lennard-Jones 6–12 potential

3.2. THE VIRIAL EXPANSION 143

The integral for the second virial coefficient is

Z ∞  n o 
σ 12 6

B2 (T ) = −2π 2
r e
4ε
− kT ( r)
−( σr )
− 1 dr. (3.2.18)
0

By making the substitution x = r/σ we cast this as

Z∞ h i
−12 −6
B2 (τ ) = −2πσ 3
x2 e−4(x −x )ε/kT − 1 dx. (3.2.19)
0

This is instructive. The σ-dependence is all in the hard core pre-factor and
the integral depends on temperature solely through the combination kT /ε.
It is possible to express the integral of Eq. (3.2.19) in terms of a Hermite
function Hn (x)1 . In this way we obtain:
 r 
2 3 √  ε 1/4 ε
B2 (T ) = πσ 2π H1 − . (3.2.20)
3 kT 2 kT
This is an elegant closed-form expression for the second virial coefficient of
the Lennard-Jones gas.
This is plotted in Fig. 3.6 together with the data from nitrogen. The
curve has been fitted with parameters ε/k = 95.5 K, σ = 3.76 Å. The figure
also shows the “hard sphere” asymptote B2hs = 67.00 cm3 /mol. The fit is
good. We see that the Lennard-Jones calculated form shows reduction in B2
at higher temperatures, where the energetic collisions can cause the atoms
to come even closer together; the “hard core” is not so hard. This effect
is observed in helium, shown in Fig. 3.11. We find, from (3.2.20), that the
maximum value, B2max = 0.529 × 23 πσ 3 occurs at T = 25.13 ε/k.
From the zero of the Hermite function we find the Boyle temperature to
be
TB = 3.418ε/k. (3.2.21)
At high temperatures we have the expansion 2

2 3 2π  ε 1/4 π  ε 3/4
B2 (T ) = πσ −
3 Γ(1/4) kT Γ(3/4) kT
2π  ε  5/4 π  
ε 7/4 5π  ε 9/4
− − − + ∙∙∙
Γ(5/4) kT 8Γ(7/4) kT 64Γ(9/4) kT
(3.2.22)
1
The Hermite polynomials Hn (x) should be familiar, from the quantum mechanics of
the harmonic oscillator. For integer order n, Hn (x) is a polynomial of power n in x.
Hermite functions generalize to the case of non-integer order; such functions are no longer
finite polynomials. We follow the Mathematica definition and terminology whereby the
same symbol is used for both: HermiteH[n, x].
2
This series is obtained by a direct term-by-term expansion of the Hermite function.
144 CHAPTER 3. NON-IDEAL GASES

Figure 3.6: Second virial coefficient of nitrogen plotted with the Lennard-
Jones functional form, Eq. (3.2.20). Lennard-Jones parameters ε/k = 95.5 K,
σ = 3.76 Å.

or, in closed form3

∞    (2n+1)/4
2 3X 1 2n − 1 4ε
B2 (T ) = − πσ Γ . (3.2.23)
3 n=0
(4n)! 4 kT

At low temperatures we have the expansion 4

r   1/2  3/2  5/2
2 π kT 15 kT 945 kT
B2 (T ) = − πσ 3 ×eε/kT × + + +∙ ∙ ∙ .
3 2 ε 16 ε 512 ε
(3.2.24)

A comment on scaling
The Lennard-Jones potential has two parameters: an energy ε and a length σ.
We note that these happen to correspond to the vertical and the horizontal
axes of the plot of U (r) against r. This means that the Lennard-Jones
potential has the form of a universal function that just needs the appropriate
scaling in the U and r directions. And by extension this tells us that for
any system of particles which interact with a Lennard-Jones potential, those
properties that depend on the inter-particle potential, similarly, will have a
3
To obtain this expression Eq. (3.2.19) is integrated by parts, the exponential is then
expanded and the integration performed term by term.
4
This is quoted from a calculation by Gutierrez [3].
3.2. THE VIRIAL EXPANSION 145

universal form when the energies are scaled by ε, the distances by σ and
other variables in the corresponding way.
It follows that any inter-particle potential which has only two adjustable
(system-specific) parameters with different dimensions, will possess this scal-
ing property. A special case of this is the hard sphere interaction which
has only one parameter; we may regard this as having an energy parame-
ter of zero. But we recognize immediately that the square well potential,
with three parameters ε, σ and the dimensionless distance ratio R does not
possess the scaling property. But if R were to be regarded as fixed then
the two parameters ε and σ would lead to universal behaviour. Indeed for
the inert gases neon, argon, krypton and xenon the values of R are close:
approximately 1.65.
It is clear that two-parameter potentials, and their resultant universal
system properties are particularly convenient in statistical mechanics. This
is one of the reasons for the popularity of the Lennard-Jones function where
the dipolar attraction and the electron shell repulsion – two very different
phenomena – are parameterized in similar ways: both having energies scaling
with the same ε and distances scaling with the same σ.
And in this vein the Sutherland potential of the next session and the
“soft sphere” interaction treated in Problem 3.16 both possess the scaling
property.

3.2.5 The Sutherland potential

The interaction between atoms or molecules comprises a repulsive part at
short distances and an attractive part at large distances. The Lennard-Jones
potential of the previous section is often used as an analytical representation
of the interaction. As we explained, the attractive tail is well-described by
the r−6 law, while the r−12 description of the repulsive core is but a simple
approximation to the actual short-range interaction. The popularity of the
6–12 potential lies principally in its mathematical elegance.
The Sutherland potential treats the short-distance repulsion in a different
way; it approximates the interaction as a hard core. The attractive tail is
described by the conventional dipolar r−6 law.
The form of the Sutherland potential is shown in Fig. 3.7; it is specified by
U (r) = ∞ r<σ
 σ 6 (3.2.25)
= −ε r > σ.
r
As with the Lennard-Jones potential, the Sutherland potential has a universal
form, scaled vertically with an energy parameter ε and horizontally with a
146 CHAPTER 3. NON-IDEAL GASES

Figure 3.7: Sutherland potential

distance parameter σ.
The second virial coefficient is given by
Z∞


B2 (T ) = −2π r2 e−U (r)/kT − 1 dr
0

so using the mathematical form for U (r), the integral splits into two parts
Zσ Z∞  ε σ 6 
B2 (T ) = 2π r2 dr − 2π r2 e kT ( r ) − 1 dr
0 σ
(3.2.26)
Z∞  
2 σ 6
r2 e ( ) − 1 dr .
ε
= πσ 3 − 2π kT r
3
σ

We substitute x = r/σ, so that

 
 Z∞   
2 ε −6
B2 (T ) = πσ 3 1 − 3 x2 e kT x − 1 dx . (3.2.27)
3  
1

This may be expressed analytically, in terms of the imaginary error function

erfi5 :  r r 
2 3 ε/kT √ ε ε
B2 (T ) = πσ e − π erfi (3.2.28)
3 kT kT
5
The standard error function erf(z) is an area under the Gaussian distribution function:
Rz 2
erf(z) = √2π 0 e−t dt. The imaginary error function is defined as erfi(z) = erf(iz)/i. The
Mathematica symbols for these are Erf[z] and Erfi[z]. We note that erfi(z) is real for
real z.
3.2. THE VIRIAL EXPANSION 147

This is plotted in Fig. 3.8 together with data from nitrogen. The curve has
been fitted with parameters ε/k = 274.2 K, σ = 3.16 Å. The figure also shows
the hard sphere asymptote B2hs = 39.81 cm3 /mol. Observe that this is not
such a good fit to the data.

Figure 3.8: Second virial coefficient of nitrogen plotted with the Sutherland
functional form, Eq. (3.2.20). Sutherland parameters ε/k = 274.2 K, σ =
3.16 Å.
The Boyle temperature for the Sutherland gas is
TB = 1.171ε/k. (3.2.29)
At high temperatures we have the series expansion
∞
X (kT /ε)−n
2
B2 (T ) = − πσ 3
3 n! (2n − 1)
 n=0 (3.2.30)
2 ε 1  ε 2 1  ε 3
= πσ 3 1 − − − − ... .
3 kT 6 kT 30 kT
while at low temperatures we have
(  2  3
2 3 ε/kT 1 kT 3 kT 15 kT
B2 (T ) = − πσ e + + + ∙∙∙ (3.2.31)
3 2 ε 4 ε 8 ε

The interesting point about the Sutherland potential is that it gives the
high-temperature behaviour of the B2 (T ) as
2  ε 
B2 (T ) ∼ πσ 3 1 − − ... ; (3.2.32)
3 kT
148 CHAPTER 3. NON-IDEAL GASES

the limiting value at high temperatures is the hard core 2πσ 3 /3, while the
leading deviation goes as T −1 .
[Compare with square well potential:
 
2 3 (R3 − 1) ε
B2 (T ) ∼ πσ 1 − − ... . (3.2.33)
3 kT
Here also the limiting high temperature value is the hard core expression
and the leading deviation goes as T −1 . Note R is dimensionless, greater than
unity. And ε is different in the two cases, i.e.
εS = (R3 − 1)εsw . (3.2.34)
]
By contrast, the second virial coefficient for the Lennard-Jones gas does
not have such a simple high-temperature behaviour – a consequence of the
“softness” of the hard core. In the high temperature limit
2 2π  ε 1/4
B2 (T ) ∼ πσ 3 ×
3 Γ(1/4) kT
 ε 1/4 (3.2.35)
2 3
∼ πσ × 1.73 ,
3 kT
so that in this case B2 (t) → 0 as T → ∞; the van der Waals second virial
coefficient tends to zero rather than the hard core limiting value.

3.2.6 Comparison of models

We plot the nitrogen second virial coefficient again, in Fig. 3.18, now showing
the best fit curves corresponding to the square well, the Lennard-Jones and
the Sutherland potentials. It will be observed that there is not much to choose
between them. The Lennard-Jones is better than the Sutherland potential
– clearly the latter is too crude. However one sees that the square well
potential provides the best fit. But one should not conclude that the square
well potential is the best physical model. Its better mathematical fit is simply
a consequence of the fact that that model has three adjustable parameters,
as compared with the two adjustable parameters of the Lennard-Jones and
the Sutherland models.
Nevertheless, from the practical perspective the square well expression
for B2 of Eq. (3.2.15) is sufficiently accurate that databases of second virial
coefficients often simply give the equation’s three parameters A, B and ε,
rather than extensive tables 6 .
6
See, for instance, the NPL Kaye and Laby web site, in particular the page
http://www.kayelaby.npl.co.uk/chemistry/3 5/3 5.html
3.2. THE VIRIAL EXPANSION 149

Figure 3.9: Second virial coefficient of nitrogen compared with fits corre-
sponding to the square well, Lennard-Jones and Sutherland potentials.

The inter-particle potentials corresponding to the fits of the various mod-

els through the nitrogen B2 data are shown in Fig 3.10.

3.2.7 Universal behaviour

The second virial coefficient B2 (T ) was obtained as an integral over the in-
teraction energy U (r) in Eq. (3.2.2). Different gases will have different values
for B2 (T ) because they will have different interaction energies.
We shall make the assumption that the interaction energy may be ex-
pressed in the following way:

U (r) = εu(r/σ). (3.2.36)

What we are saying here is that U (r) has a universal shape and, for a given
gas, it is scaled vertically by a specific energy ε and horizontally by a specific
length σ. We note that the Lennard-Jones interaction is of this form and
that values of ε and σ are tabulated, see for example Kittel [4].
Using this expression for U (r) in Eq. (3.2.2), by a change of variable we
obtain Z ∞
 
3
B2 /σ = −2π x2 e−u(x)/τ − 1 dx (3.2.37)
0

where τ = kT /ε is a reduced temperature. Then to the extent that our

scaling assumption for U (r) is valid, plots of B2 /σ 3 against kT /ε would
150 CHAPTER 3. NON-IDEAL GASES

Figure 3.10: Square well, Lennard-Jones and Sutherland potentials corre-

sponding to Nitrogen second virial coefficient fits.

fall on a single curve. In Fig. 3.11 we show such data for the noble gases,
scaled with their Lennard-Jones ε and σ. Apart from helium, the points do
indeed collapse onto a single curve. This is provides support for our scaling
assumption; this will be formalized into the Law of Corresponding States
in the next chapter. The points in Fig. 3.11 fall very close to the curve
calculated from the Lennard-Jones potential, Eq. (3.2.20).
Helium is special: because of its light mass, quantum effects are impor-
tant. This is discussed in the next section.

3.2.8 Quantum gases – the special case(s) of helium

The main aim of this section is to explain why helium’s second virial coef-
ficient points, in Fig. 3.11, do not fall on the universal curve, which follows
from the scaling property of the interaction potential. But first we shall see
that a quantum gas has a non-zero B2 even in the absence of any interactions.

Non-interacting quantum gas

Now the second virial coefficient is a measure of the deviation from the ideal
gas equation of state. But at low temperatures the equation of state for a
non-interacting gas of quantum particles deviates from the ideal gas equation
of state. We saw this in Section 2.7.1. This deviation, when expressed as a
series in powers of density may be regarded as a virial expansion. This is
3.2. THE VIRIAL EXPANSION 151

Figure 3.11: Reduced second virial coefficient of the Noble Gases, together
with Lennard-Jones form. The helium data are for the abundant isotope
4
He.

precisely what we did in Eq. (2.7.24), which we shall write here as

( √  2 )
π 3/2 ~3 N π 3 16 3 − 27 ~6 N
pV = N kT 1 ± − 2 + ...
2α (mkT )3/2 V α 27 (mkT )3 V
(3.2.38)
where the + is for fermions and the − is for bosons, and α is the spin
degeneracy factor. From this expression we can directly read off the second
and third virial coefficients. In particular

π 3/2 ~3
B20 (T ) = ±
2α (mkT )3/2 (3.2.39)
1
= ± 5/2 Λ3 (T )
2 α
where Λ is the thermal de Broglie wavelength, Eq. (2.3.5), and the “0” su-
perscript indicates this is in the absence of any interaction potential. B2 is
monotonic in T ; it is positive for fermions and negative for bosons. This is
plotted in Fig. 3.12.
This result has an interesting interpretation. Recall the hard sphere ex-
pression for B2 was essentially the volume of a particle. Here we see that
this quantum contribution to B2 is essentially the thermal deBroglie volume
of the particle – positive for fermions and negative for bosons.
152 CHAPTER 3. NON-IDEAL GASES

Figure 3.12: Second virial coefficient of a quantum ideal gas

It is often stated crudely that requirements of quantum statistics leads

to “exchange forces” on classical “Maxwell” particles; the exclusion principle
for fermions gives a repulsive force while bosons experience an attraction.
Such a view can, however, be seriously misleading [5].

Interacting quantum gas

Quantum mechanics may be regarded as influencing the second virial coeffi-

cient of a gas in two different ways. Particles are delocalized over a length
scale Λ, and particle “statistics” will determine the symmetry of the states
that are included in the partition function sum. The Lee-Yang approach to
quantum statistical Mechanics [6] allows one to treat these contributions as
separate and additive. We shall therefore write

B2 (T ) = B2d (T ) + B2ex (T ) (3.2.40)

where B2d (T ) is the delocalization or “direct” term and B2ex (T ) is the statistics
or “exchange” term.
Of course a full quantum treatment will incorporate both contributions,
but such calculations are rather complex and tedious [2]. Instead, we shall
examine the two contributions separately. This will provide an intuitive un-
derstanding of the way quantum effects influence the second virial coefficient
of the helium gases. We shall denote these two different contributions as the
statistics and the delocalization contributions.
3.2. THE VIRIAL EXPANSION 153

Statistics contribution to B2
For non-interacting fermions or bosons there is only the statistics contribution
to B2 and this is given by Eq. (3.2.39). However in the presence of interactions
this is dramatically attenuated [7], so that the overwhelming effect is from
the delocalization.

Delocalization contribution to B2
Quantum mechanics may be regarded as causing a delocalization of the
atomic positions and this may be accommodated by a renormalization of
the inter-atomic interaction. This idea was suggested by Feynman [8] and
subsequently the procedure was developed by Young [9] for the case of the
Lennard-Jones interaction. Essentially one averages the interaction potential
over a Gaussian probability density whose width is the thermal de Broglie
wavelength. The result is a Lennard-Jones interaction with renormalized ε
and σ which depend on temperature. We shall simply quote the result, and
refer the interested reader to the original references for the details.
In terms of the Lennard-Jones parameters ε and σ, we define a charac-
teristic temperature T ∗ : r
∗ 1 ~ ε
T = (3.2.41)
6π kσ m
in terms of which we can introduce a reduced temperature variable τ
τ = T /T ∗ . (3.2.42)
Then the renormalization of the Lennard-Jones ε and σ is given, in terms of
the reduced temperature, by:
ε → εˉ(τ ) = E(τ )ε
(3.2.43)
σ →σ ˉ (τ ) = S(τ )σ
where
 −3/4
E(τ ) = 1 + 19.1τ −1 + f (τ )τ −2
 1/2 (3.2.44)
S(τ ) = 1 + g(τ )τ −1 .
and
 
f (τ ) = 5 + (177.7 − 5) 1 − e−τ /250
  (3.2.45)
g(τ ) = 4 + (7.54 − 4) 1 − e−τ /250
These quantum renormalization factors are shown, as a function of reduced
temperature τ in Fig. 3.14
154 CHAPTER 3. NON-IDEAL GASES

Figure 3.13: Quantum renormalization factors for the Lennard-Jones σ and

ε parameters.

Consequences for Helium

There are two isotopes of helium: the abundant and heavier 4 He and the
rarer and lighter 3 He. We note that 4 He is a boson and 3 He is a fermion. I
have plotted B2 data for the helium isotopes in Fig. 3.14 together with the
Lennard-Jones curve, as in Fig. 3.11.
While the Lennard-Jones parameters for the two isotopes will be the
same, since they have different masses, the renormalization factors will be
different. By applying these renormalizations to the ε and σ in Eq. (3.2.20)
we obtain the two dashed curves in Fig. 3.14. The bare Lennard-Jones curve
has been shifted almost through the data points.
From this we conclude that delocalization effect almost accounts for the
second virial coefficients of 3 He and 4 He. The difference between B2 for
the two isotopes is predominantly because of their different masses and not
because of their different statistics.

3.3 Thermodynamics
3.3.1 Throttling
In a throttling process a gas is forced through a flow impedance such as a
porous plug. For a continuous process, in the steady state, the pressure will
be constant (but different) either side of the impedance. When this happens
to a thermally isolated system so that heat neither enters nor leaves the sys-
3.3. THERMODYNAMICS 155

Figure 3.14: Second virial coefficient of 3 He and 4 He.

tem then it is referred to as a Joule-Kelvin or Joule Thompson process. This

is fundamentally an irreversible process, but the arguments of thermodynam-
ics are applied to such a system simply by considering the equilibrium initial
state and the equilibrium final state which applied way before and way after
the actual process. This throttling process may be modelled by the diagram
in Fig. 3.15.

Figure 3.15: Joule-Kelvin throttling process

Work must be done to force the gas through the plug. The work done is
Z 0 Z V2
ΔW = − p1 dV − p2 dV = p1 V1 − p2 V2 . (3.3.1)
V1 0

Since the system is thermally isolated the change in the internal energy is
due entirely to the work done:
E 2 − E 1 = p 1 V 1 − p2 V 2 (3.3.2)
or
E 1 + p 1 V 1 = E2 + p2 V 2 . (3.3.3)
156 CHAPTER 3. NON-IDEAL GASES

The enthalpy H is defined by

H = E + pV (3.3.4)

thus we conclude that in a Joule-Kelvin process the enthalpy is conserved.

The interest in the throttling process is that whereas for an ideal gas
the temperature remains constant, it is possible to have either cooling or
warming when the process happens to a non-ideal gas. The operation of
most refrigerators is based on this.

3.3.2 Joule-Thomson coefficient

The fundamental differential relation for the enthalpy is

dH = T dS + V dp. (3.3.5)

It is, however, rather more convenient to use T and p as the independent

variables rather than the natural S and p. This is effected by expressing the
entropy as a function of T and p whereupon its differential may be expressed
as  
∂S  ∂S 
dS = dT + dp. (3.3.6)
∂T p ∂p T
But 
∂S  cp
 = (3.3.7)
∂T p T
and using a Maxwell relation we have
 
∂S  ∂V 
=− (3.3.8)
∂p T ∂T p

so that ( )
∂V 
dH = cp dT + V −T dp. (3.3.9)
∂T p
Now since H is conserved in the throttling process dH = 0 so that
(  )
1 ∂V 
dT = T − V dp (3.3.10)
cp ∂T p

which tells us how the temperature change is determined by the pressure

change. The Joule-Thomson coefficient μJ is defined as the derivative

∂T 
μJ = , (3.3.11)
∂p H
3.3. THERMODYNAMICS 157

giving ( )

1 ∂V 
μJ = T −V (3.3.12)
cp ∂T p
This is zero for the ideal gas (Problem 3.1). When μJ is positive then the
temperature decreases in a throttling process when a gas is forced through a
porous plug.

Figure 3.16: Isenthalps and inversion curve for nitrogen (after Zemansky[10])

3.3.3 Connection with the second virial coefficient

We consider the case where the second virial coefficient gives a good approx-
imation to the equation of state. Then we are assuming that the density is
low enough so that the third and higher coefficients can be ignored. This
means that the second virial coefficient correction to the ideal gas equation
is small and then solving for V in the limit of small B 2 (T ) gives
N kT
V = + N B2 (T ) . (3.3.13)
p
so that the Joule-Thomson coefficient is then
 
N T dB2 (T ) B2 (T )
μJ = − . (3.3.14)
cp dT T
158 CHAPTER 3. NON-IDEAL GASES

Within the low density approximation it is appropriate to use the ideal gas
thermal capacity
5
cp = N k (3.3.15)
2
so that  
2T dB2 (T ) B2 (T )
μJ = − . (3.3.16)
5k dT T

3.3.4 Inversion temperature

The behaviour of the Joule-Thomson coefficient can be seen from the follow-
ing construction. We take the shape of B 2 (T ) from the square well potential
model. While not qualitatively correct, this does exhibit the general features
of a realistic interparticle potential.

Figure 3.17: Behaviour of the Joule-Thomson coefficient

We see that at low temperatures the slope of the curve, dB/dT is greater
than B/T so that μJ is positive, while at high temperatures the slope of
the curve, dB/dT is less than B/T so that μJ is negative. The temperature
where μJ changes sign is called the inversion temperature, Ti .
The inversion curve for nitrogen is shown as the dashed line in Fig. 3.16.
We see that at high temperatures μJ is negative, as expected. As the tem-
perature is decreased the inversion curve is crossed and μJ becomes positive.
Note, however that the the low density approximation, implicit in going only
to the second virial coefficient, keeps us away from the lower temperature
region where the gas is close to condensing, where the Joule-Thomson coef-
ficient changes sign again.
3.4. VAN DER WAALS EQUATION OF STATE 159

Sutherland bit
The Boyle temperature and the inversion temperature for this gas may be
found from their definitions
B2 (T ) = 0 → TB
dB2 (T ) B2 (T ) (3.3.17)
− =0 → Ti
dT T
to give

TB = 1.171ε/k
(3.3.18)
Ti = 2.215ε/k.

The tangent construction for the inversion temperature (Section 3.3.4 and
Fig. 3.8) is shown in Fig. 3.18. The ratio is then

Figure 3.18: Boyle temperature and inversion temperature

Ti /TB = 1.259. (3.3.19)

3.4 Van der Waals Equation of State

3.4.1 Approximating the Partition Function
Rather than perform an exact calculation as a series in powers of an expansion
parameter – the density or the cluster function fij – in this section we shall
adopt a different approach by making an approximation to the partition
160 CHAPTER 3. NON-IDEAL GASES

function, which should be reasonably valid at all densities. Furthermore the

approximation we shall develop will be based on the single-particle partition
function. We shall, in this way, obtain an equation of state that approximates
the behaviour of real gases. This equation was originally proposed by van der
Waals in his Ph. D. Thesis in 1873. An English translation is available [11]
and it is highly readable; van der Waals’s brilliance shines out.
In the absence of an interaction potential the single-particle partition
function is
V
z = 3. (3.4.1)
Λ
Recall that the factor V here arises from integration over the position coordi-
nates. The question now is how to account for the inter-particle interactions
– in an approximate way. Now the interaction U (r) comprises a strong re-
pulsive hard core at short separations and a weak attractive long tail at large
separations. And the key is to treat these two parts of the interaction in
separate ways.
• The repulsive core effectively excludes regions of space from the integra-
tion over position coordinates. This may be accounted for by replacing
V by V − Vex where Vex is the volume excluded by the repulsive core.
• The attractive long tail is accounted for by including a factor in the
expression for z of the form
e−hEi/kT (3.4.2)
where hEi is an average of the attractive part of the potential.
Thus we arrive at the approximation
V − Vex −hEi/kT
z= e . (3.4.3)
Λ3
Note that we have approximated the interaction by a mean field assumed to
apply to individual particles. This allows us to keep the simplifying feature
of the free-particle calculation where the many-particle partition function
factorises into a product of single-particle partition functions. This is ac-
cordingly referred to as a mean field calculation.

3.4.2 Van der Waals Equation

The equation of state is found by differentiating the free energy expression:
 
∂ ln Z  ∂ ln z 
p = kT = N kT . (3.4.4)
∂V T,N ∂V T
3.4. VAN DER WAALS EQUATION OF STATE 161

Now the logarithm of z is

ln z = ln (V − Vex ) − 3 ln Λ − hEi /kT (3.4.5)

so that 
∂ ln z  N kT d hEi
p = N kT  = −N (3.4.6)
∂V T V − Vex dV
since we allow the average interaction energy to depend on volume (density).
This equation may be rearranged as

d hEi N kT
p+N = (3.4.7)
dV V − Vex
or  
d hEi
p+N (V − Vex ) = N kT. (3.4.8)
dV
This is similar to the ideal gas equation except that the pressure is increased
and the volume decreased from the ideal gas values. These are constant
parameters. They account, respectively, for the attractive long tail and the
repulsive hard core in the interaction. Conventionally we express the param-
eters as aN 2 /V 2 and N b, so that the equation of state is
 
N2
p + a 2 (V − N b) = N kT (3.4.9)
V

and this is known as the van der Waals equation.

Some isotherms of the van der Waals equation are plotted in Fig. 3.19
for three temperatures T1 > T2 > T3 . On the right hand side of the plot,
corresponding to low density, we have gaseous behaviour; here the van der
Waals equation gives small deviations from the ideal gas behaviour. On
the left hand side, particularly at the lower temperatures, the steep slope
indicates incompressibility. This is indicative of liquid behaviour. The non-
monotonic behaviour at low temperature is peculiar and indeed it is non-
physical. This will be discussed in detail in the next chapter.
The van der Waals equation gives a good description of the behaviour of
both gases and liquids. In introducing this equation of state we said that the
method should treat both low-density and high-density behaviour, and this
it has done admirably. For this reason Landau and Lifshitz [12] refer to the
van der Waals equation as an interpolation equation. The great power of the
equation, however, is that it also gives a good qualitative description of the
gas-liquid phase transition, to be discussed in Chapter 4.
162 CHAPTER 3. NON-IDEAL GASES

Figure 3.19: van der Waals isotherms

3.4.3 Estimation of van der Waals Parameters

In the van der Waals approach the repulsive and the attractive parts of the
inter-particle interaction were treated separately. Within this spirit let us
consider how the two parameters of the van der Waals equation might be
related to the two parameters of the Lennard-Jones inter-particle interaction
potential. The repulsion is strong; particles are correlated when they are very
close together. We accounted for this by saying that there is zero probability
of two particles being closer together than σ. Then, as in the hard core
discussion of Section 3.2.1, the region of co-ordinate space is excluded, and
the form of the potential in the excluded region (U (r) very large) does not
enter the discussion. Thus just as in the discussion of the hard core model,
the excluded volume will be
2
Vex = N πσ 3 . (3.4.10)
3
The attractive part of the potential is weak. Here there is very little
correlation between the positions of the particles; we therefore treat their
distribution as approximately uniform. The mean interaction for a single
pair of particles hEp i is then
Z
1 ∞
hEp i = 4πr 2 U (r) dr
V σ
Z   
1 ∞ 2 σ 12  σ 6
= 4πr 4ε − dr (3.4.11)
V σ r r
32πσ 3
=− ε.
9V
3.4. VAN DER WAALS EQUATION OF STATE 163

Now there are N (N −1)/2 pairs, each interacting through U (r), so neglecting
the 1, the total energy of interaction is N 2 hEp i /2. This is shared among the
N particles, so the mean energy per particle is
hEi = hEp i N /2
16πσ 3 N (3.4.12)
=− ε.
9 V
In the van der Waals equation it is the derivative of this quantity we require.
Thus we find  2
d hEi 16 3 N
N = πσ ε. (3.4.13)
dV 9 V
These results give the correct assumed N and V dependence of the param-
eters used in the previous section. So finally we identify the van der Waals
parameters a and b as
16 3
a= πσ ε
9 (3.4.14)
2
b = πσ 3 .
3

3.4.4 Virial Expansion

It is a straightforward matter to expand the van der Waals equation as a
virial series. We express p/kT as
p N aN 2
= −
kT V − N b kT V 2
  −1  2 (3.4.15)
N N a N
= 1−b −
V V kT V
and this may be expanded in powers of N/V to give
   2   3  4
p N N a  N 2 N
= + b− + b + b3 + . . . . (3.4.16)
kT V V kT V V
Thus we immediately identify the second virial coefficient as
a
B2VW (T ) = b − . (3.4.17)
kT
This has the form as sketched for the square well potential. For this model
we can find the Boyle temperature and the inversion temperature:
a
TB = ,
bk
2a (3.4.18)
Ti = .
bk
164 CHAPTER 3. NON-IDEAL GASES

So we conclude that for the van der Waals gas the inversion temperature is
double the Boyle temperature.
Incidentally, we observe that the third and all higher virial coefficients,
within the van der Waals model, are constants independent of temperature.

3.5 Other Phenomenological Equations of State

3.5.1 The Dieterici equation
The Dieterici equation of state is one of a number of phenomenological equa-
tions crafted to give reasonable agreement with the behaviour of real gases.
The interest in the Dieterici is twofold.
1. The equation gives a better description of the behaviour of fluids in
the vicinity of the critical point than does the van der Waals equation.
This will be discussed in Chapter 4, in Section 4.2.

2. The equation is consistent with the Third Law of Thermodynamics [13].

The Dieterici equation may be written as
Na
p (V − N b) = N kT e− kT V . (3.5.1)

As with the van der Waals equation, this equation has two parameters, a
and b, that parameterise the deviation from ideal gas behaviour.
For the present we briefly examine the virial expansion of the Dieterici
equation. In other words we will look at the way this equation treats the
initial deviations from the ideal gas.

Virial expansion
In order to obtain the virial expansion we express the Dieterici equation as
p N Na
= e− kT V . (3.5.2)
kT V − Nb
And from this we may expand to give the series in N /V
 2   3  
p N N a  N 2 a2 ab
= + b− + b − 2 2− + . . . (3.5.3)
kT V V kT V 2k T kT
This gives the second virial coefficient to be
a
B2D = b − . (3.5.4)
kT
3.5. OTHER PHENOMENOLOGICAL EQUATIONS OF STATE 165

This is the same as that for the van der Waals gas, and the parameters
a and b may thus be identified with those of the van der Waals model.
As a consequence, we conclude that both the van der Waals gas and the
Dieterici gas have the same values for the Boyle temperature and the inversion
temperature.
The third virial coefficient is given by

a2 ab
B3D (T ) = b2 − 2 2
− ; (3.5.5)
2k T kT
we see that this depends on temperature, unlike that for the van der Waals
equation, which is temperature-independent.

3.5.2 The Berthelot equation

As with the Dieterici equation, the Berthelot equation is another of phe-
nomenological origin. The equation is given by
 
αN 2
p+ (V − N b) = N kT. (3.5.6)
kT V 2

The parameters of the Berthelot equation are given by α and b. We observe

this equation is very similar to the van der Waals equation; there is a slight
difference in the pressure-correction term that accounts for the long distance
attraction of the intermolecular potential.
Since the Berthelot and van der Waals equation are related by a = α/kT
it follows that the Berthelot second virial coefficient is given by
α
B2B = b − . (3.5.7)
(kT )2

3.5.3 The Redlich-Kwong equation

Most improved phenomenological equations of state involve the introduc-
tion of additional parameters. The Redlich-Kwong equation is as good as
many multi-parameter equations but, as with the previous equations we have
considered, it has only two parameters. We shall write the Redlich-Kwong
equation as
N kT aN 2
p= −√ . (3.5.8)
V − Nb kT V (V + N b)
This should be compared to the similar expression for the van der Waals
equation, Eq. (3.4.15).
166 CHAPTER 3. NON-IDEAL GASES

The virial expansion is

 2    3  
p N N a N 2 ab
= + b− + b + + . . . (3.5.9)
kT V V (kT )3/2 V (kT )3/2
so, in particular, we identify the Redlich-Kwong second virial coefficient to
be
a
B2RK = b − (3.5.10)
(kT )3/2
(note: kTc = 0.345(a/b)2/3 , Vc = 3.847N b, pc = 0.02989a2/3 /b5/3 )

3.6 Hard Sphere Gas

The interactions between the atoms or molecules of a real gas comprise a
strong repulsion at short distances and a weak attraction at long distances.
Both of these are important in determining how the properties of the gas
differ from those of an ideal (non-interacting) gas. We have already consid-
ered various approximations to the inter-particle interaction when we looked
at initial deviations from ideal behaviour in the calculations of the virial
expansions in Section 3.2. We considered a sequence of model approxima-
tions from the simplest: the hard core potential, to the most realistic: the
Lennard-Jones 6-12 potential.
In this section we shall return for a deeper study of the to the hard core
potential. In justification we can do no better than to quote from Chaikin
and Lubensky[14] p.40: “Although this seems like an immense trivialisation
of the problem, there is a good deal of unusual and unexpected physics to be
found in hard-sphere models.”
We recall that the hard core potential, Eq. (3.2.3) is
U (r) = ∞ r<σ
=0 r>σ
where σ is the hard core diameter. This is indeed a simplification of a real
inter-particle interaction – but what behaviour does it predict? What prop-
erties of real systems can be understood in terms of the short-distance repul-
sion? And, indeed, what properties cannot be understood from this simpli-
fication?

3.6.1 Possible approaches

The direct way of solving the problem of the hard sphere fluid would be to
evaluate the partition function; everything follows from that. Even for an
3.6. HARD SPHERE GAS 167

interaction as simple as this, it turns out that the partition function cannot
be evaluated analytically except in one dimension; this is the so-called Tonks
hard stick model, which leads to the (one-dimensional) Clausius equation of
state
p(L − Lex ) = N kT. (3.6.1)

Certainly in two and three dimensions no explicit solution is possible. 7 Argu-

ments about why the partition function (really the configuration integral) is
so difficult to evaluate are given in Reif [15]. The point is that the excluded
volumes appear in nested integrals and these are impossible untangle, except
in one dimension. Accepting that no analytic solution is possible in three
dimensions, there is a number of approaches that might be considered.

• Mean field – this will indicate the general behaviour to be expected,

• Virial expansion – this represents first-principles theoretical calculation.

• Molecular dynamics – this may be viewed as accurate “measurements

made by computer”.

The mean field approach 8 to the the hard sphere gas results in the Clausius
equation of state: the ideal gas equation, but with an excluded volume term.

p(V − Vex ) = N kT. (3.6.2)

This follows by analogy with our treatment of the van der Waals gas, where
now there is no attractive term in the interaction. See also Problem 3.8.
There are extensive molecular dynamics simulations, see in particular
Bannerman et al.[16] and it is even possible to do your own; the applications
of Gould and Tobochnik [17] are very instructive for this.
We shall look at virial expansions and see how far they may be “pushed”.
In other words our interest is in what analytical conclusions may be draw.
We can then compare these conclusions with the molecular dynamics “ex-
perimental data”.
7
(Question: is “excluded volume” treatment a mean-field treatment – and so is the
excluded volume argument then valid for four and higher dimensions? This can be tested
using the virial coefficients calculated by Clisby and McCoy for four and higher dimensions.
– The answer is NO; strictly speaking “excluded volume” is not part of the mean field
procedure)
8
But see footnote above – this is not truly a mean field procedure.
168 CHAPTER 3. NON-IDEAL GASES

3.6.2 Hard Sphere Equation of state

The equation of state of a hard-sphere fluid has a very special form. Recall
that the Helmholtz free energy F is given in terms of the partition function
Z by
F = −kT ln Z. (3.6.3)
We saw that the partition function for an interacting gas may be written as
Z = Zid Q (3.6.4)
where Zid is the partition function for an ideal (non-interacting) gas
 N
1 V
Z= (3.6.5)
N ! Λ3
and Q is the configuration integral
Z P
1
Q= N e−( i<j U (qi ,qj ))/kT d3N q. (3.6.6)
V
To obtain the equation of state we must find the pressure, by differentiating
the free energy

∂F 
p=−
∂V T,N

∂ ln Z 
= kT (3.6.7)
∂V T,N
  !
∂ ln Zid  ∂ ln Q 
= kT + .
∂V T,N ∂V T,N
It is important, now, to appreciate that the configuration integral is indepen-
dent of temperature. This must be so, since there is no energy scale for the
problem; the interaction energy is either zero or it is infinite. Thus the ratio
E/kT will be temperature-independent. The pressure of the hard-sphere gas
must then take the form
 
N
p = kT + g(N/V ) . (3.6.8)
V
The function g(N/V ) is found by differentiating ln Q with respect to V .
We know it is a function of N and V and in the thermodynamic limit the
argument must be intensive. Thus the functional form, and we have the
low-density ideal gas limiting value g(0) = 0.
The important conclusion we draw from these arguments, and in partic-
ular from Eq. (3.6.8) is that for a hard sphere gas the combination p/kT is
a function of the density N/V . This function must depend also on the only
parameter of the interaction: the hard core diameter σ.
3.6. HARD SPHERE GAS 169

3.6.3 Virial Expansion

The virial expansion, Eq. (3.2.1), is written as
 2  3
p N N N
= + B2 + B3 + .... (3.6.9)
kT V V V
where the virial coefficients Bm are, in the general case, functions of tempera-
ture. However, as argued above, for the hard sphere gas the virial coefficients
are temperature-independent.
The virial expansion may be regarded as a low-density approximation to
the equation of state. Certainly this is the case when only a finite number
of coefficients is available. If, however, all the coefficients were known, then
provided the series were convergent, the sum would give p/kT for all values of
the density N/V up to the radius of convergence of the series : the complete
equation of state. Now although we are likely to know the values for but
a finite number of the virial coefficients, there may be ways of guessing /
inferring / estimating the higher-order coefficients. We shall examine two
ways of doing this.

3.6.4 Virial Coefficients

The second virial coefficient for the hard sphere gas was calculated in Sec-
tion 3.2.2; we found
2
B2 = πσ 3 , (3.6.10)
3
independent of temperature, as expected.
The general term of the virial expansion is ( N
V
)m Bm , which must have the
dimensions of N/V . Thus Bm will have the dimensions of V m−1 . Now the
only variable that the hard sphere virial coefficients depend on is σ. Thus it
is clear that
Bm = const × σ 3(m−1) (3.6.11)
where the constants are dimensionless numbers – which must be determined.
It is increasingly difficult to calculate the higher-order virial coefficients;
the second and third were calculated by Boltzmann in 1899; those up to
sixth order were evaluated by Ree and Hoover[18] in 1964, and terms up to
tenth order were found by Clisby and McCoy[19] in 2006. More recently the
eleventh and twelfth were obtained by Wheatley in 2013[20]. These all are
listed in the table below, in terms of the single parameter b:
2
b = B2 = πσ 3 . (3.6.12)
3
170 CHAPTER 3. NON-IDEAL GASES

B2 /b = 1
B3 /b2 = 0.625
B4 /b3 = 0.28694950
B5 /b4 = 0.11025210
B6 /b5 = 0.03888198
B7 /b6 = 0.01302354
B8 /b7 = 0.00418320
B9 /b8 = 0.00130940
B10 /b9 = 0.00040350
B11 /b10 = 0.00012300
B12 /b11 = 0.00003700

Table 3.1: Virial coefficients for the hard sphere gas. B2 and B3 calculated
by Boltzmann, B4 to B6 by Ree and Hoover, B7 to B10 by Clisby and McCoy.

Note/recall that the hard sphere virial coefficients are independent of tem-
perature (Problem 3.7) and they are all expressed in terms of the hard core
dimension.
We now consider ways of guessing / inferring / estimating the higher-order
coefficients, so that the hard sphere equation of state may be approximated.

3.6.5 Carnahan and Starling procedure

We start with the remarkable procedure of Carnahan and Starling[21]. They
inferred a general (approximate) expression for the nth virial coefficient, en-
abling them to sum the virial expansion and thus obtain an equation of state
in closed form. The virial expansion is written as
   2
pV N N
= 1 + B2 + B3 + .... (3.6.13)
N kT V V

In 1969 only the first six virial coefficients, from Ree and Hoover, were known.
Carnahan and Starling specified the density as the fraction of the volume
occupied by the spheres. The volume of a sphere of diameter σ is 16 πσ 3 or
b/4. So the packing fraction y is given by y = N b/4V in terms of which
Carnahan and Starling wrote the virial expansion as 9

pV
= 1 + 4y + 10y 2 + 18.36y 3 + 28.22y 4 + 39.82y 5 + ∙ ∙ ∙ . (3.6.14)
N kT
9
Actually Carnahan and Starling had a slight, but insignificant error in their final
term’s coefficient; Ree and Hoover’s B6 was not quite right.
3.6. HARD SPHERE GAS 171

It is convenient to introduce “reduced” virial coefficients βn such that Car-

nahan and Starling series is
pV
= 1 + β2 y + β3 y 2 + ∙ ∙ ∙ + βn y n−1 + ∙ ∙ ∙ (3.6.15)
N kT
Here
βn = 4n−1 Bn /bn−1 (3.6.16)
and we tabulate the βn :

β2 = 4
β3 = 10
β4 = 18.364768
β5 = 28.224512
β6 = 39.81514752
β7 = 53.34441984
β8 = 68.5375488
β9 = 85.8128384
β10 = 105.775104
β11 = 128.974848
β12 = 155.189248

Table 3.2: Reduced virial coefficients for the hard sphere gas

This was the Carnahan and Starling train of argument:

• they observed that the βn coefficients were “close to” integers;
• they noted that if they rounded to whole numbers: 4, 10, 18, 28, 40,
then βn was given by (n − 1)(n + 2);
• they then made the assumption that this expression would work for the
higher-order terms as well;
• this enabled them to sum the virial series, to obtain an equation of
state in closed form.
So they had a suggestion for the values of the virial coefficients to all orders.
We can check their hypothesis, based upon the coefficients known to them,
by comparison with the newly-known virial coefficients.

n 2 3 4 5 6 7 8 9 10 11 12
rounded true βn 4 10 18 28 40 53 69 86 106 129 155
C+S: (n − 1)(n + 2) 4 10 18 28 40 54 70 88 108 130 154
172 CHAPTER 3. NON-IDEAL GASES

The agreement is not quite perfect, but it is still rather good.

The Carnahan and Starling virial series is then
pV
= 1 + 4y + 10y 2 + 18y 3 + 28y 4 + 40y 5 + ∙ ∙ ∙
N kT
X∞
=1+ (n − 1)(n + 2)y n−1 (3.6.17)
n=2
X∞
=1+ n(n + 3)y n .
n=1

The series is summed, to give

pV 2y(2 − y)
=1+
N kT (1 − y)3
(3.6.18)
1 + y + y2 − y3
= .
(1 − y)3
This is the Carnahan and Starling equation of state. In Fig. 3.20 we have
plotted this equation together with molecular dynamics simulation data from
Bannerman et al.[16]. The agreement between the Carnahan and Starling

Figure 3.20: Molecular dynamics simulation data plotted with the Carnahan
and Starling equation of state

equation of state and the molecular dynamics data is highly impressive. The
3.6. HARD SPHERE GAS 173

drop in pV /N kT at y = 0.524 is “associated” with the transition to a solid

phase.
From Carnahan and Starling’s model the general expression for the nth
virial coefficient is
(n − 1)(n + 2) n−1
Bn = b (3.6.19)
4n−1
These are listed in Table 3.6.5 together with the true values.

Calculated C+S value

B2 /b = 1 1
B3 /b2 = 0.625 0.625
B4 /b3 = 0.28694950 0.28125
B5 /b4 = 0.11025200 0.10937500
B6 /b5 = 0.03888198 0.03906250
B7 /b6 = 0.01302354 0.01318359
B8 /b7 = 0.00418320 0.00427246
B9 /b8 = 0.00130940 0.00134277
B10 /b9 = 0.00040350 0.00041199
B11 /b10 = 0.00012300 0.00012398
B12 /b11 = 0.00003700 0.00003672

Table 3.3: Carnahan and Starling’s hard sphere virial coefficients

Incidentally, the universal function g(n) of Eq. (3.6.8) is then given by

n2 b 2 − 14 nb
g(n) =
. (3.6.20)
2 1 − 1 nb 3
4

A more systematic way at arriving at an equation of state is the Padé

method.

3.6.6 Padé approximants

The equation of state of the hard sphere gas takes the form
pV
= f (y), (3.6.21)
N kT
where we are writing y = N b/4V and f is a universal function of its argument.
So if the function is determined then the hard sphere equation of state is
known.
The virial expansion gives f as a power series in its argument. And in
reality one can only know a finite number of these terms. The Carnahan and
174 CHAPTER 3. NON-IDEAL GASES

Starling procedure took the known terms, it guessed the (infinite number of)
higher-order terms and it then summed the series. The figure above indicates
that the result is good, but it relied on guesswork and intuition.
For the Carnahan and Starling equation of state the function f may be
written as
1 + y + y2 − y3
f (y) = (3.6.22)
1 − 3y + 3y 2 − y 3
In this form we observe that f (y) is the quotient of two polynomials. And
this leads us naturally to the Padé method; this is the general framework for
making approximations as such quotients.
One knows f (x) to a finite number of terms, say N . The Padé method
provides a systematic procedure for approximating the higher-order terms
and summing the series. In the Padé method the true function f (x) is ap-
proximated by the quotient of two polynomials

Pn (x)
f (x) ≈ Fnm (x) = . (3.6.23)
Qm (x)

Here Pn (x) and Qm (x) are polynomials of degrees n and m respectively:

Pn (x) = p0 + p1 x + p2 x2 + ... + pn xn ,
(3.6.24)
Qm (x) = q0 + q1 x + q2 x2 + ... + qm xm .

Without loss of generality we may (indeed it is convenient to) restrict q0 = 1.

This will ensure the coefficients pi , qj (for a given n, m) are unique.
All the coefficients of Pn (x) and Qm (x) may be determined so long as
f (x) is known to at least n + m terms. In other words if f (x) is known to
N = n + m terms then Fnm (x) agrees with the known terms of the series for
f (x). However the quotient generates a sequence of higher order terms as
well. And the hope is that this series will be a good approximation to the
true (but unknown) f (x).
The power series of
f (x)Qm (x) − P (x) (3.6.25)
begins with the term in xm+n+1 . In other words the coefficient of this and
the higher powers are “manufactured” by the Padé procedure.
One can construct approximants with different m, n subject to m+n = N .
See Reichl[2] for (some) details. Essentially there will be a subset of m, n pairs
whose Padé aproximants appear similar, usually when m ∼ n ∼ N/2. These
“robust” approximants would be expected to provide a good approximation
to the true function.
3.7. BRIDGE TO THE NEXT CHAPTER 175

In this way in 1964 Ree and Hoover[18], using the then known B2 to B6
(i.e. before Clisby and McCoy’s extra virial coefficients) constructed the 3-2
Padé approximant:
pV 1 + 1.81559y + 2.45153y 2 + 1.27735y 3
= . (3.6.26)
N kT 1 − 2.18441y + 1.18916y 2
This is plotted as the solid line in Fig. 3.21. For comparison the dotted line
shows the truncated virial series up to B6 .

Figure 3.21: The 3-2 Padé approximant (solid line) and truncated virial
series up to B6 (dashed line) shown with molecular dynamics simulations of
the hard sphere gas

Observe the 3-2 Padé gives very good agreement with the molecular dy-
namics data. By contrast the corresponding truncated virial series is essen-
tially useless; this indicates the value of the Padé procedure. We show the
original Ree-Hoover √results in Fig. 3.22. Note, however, they uses a different
density scale: VV0 = π18 y, moreover they only had four molecular dynamics
data points and there is no evidence of solidification.

3.7 Bridge to the next chapter

The intention of this chapter has been to show how inter-particle interactions
change the properties of a gas from the canonical ideal gas behaviour. In-
176 CHAPTER 3. NON-IDEAL GASES

Figure 3.22: Ree and Hoovers Padé approximation to the hard sphere equa-
tion of state

teractions were assumed to be “sufficiently weak” so that the behaviour was

still gas-like. We know that when interactions are stronger they can result
in a qualitatively different phase; thus a gas might condense into a liquid or
even into a solid. The next chapter is devoted to the study of phase transi-
tions. It follows logically; this chapter was concerned with the weaker effects
of interactions and the next chapter with stronger effects.
There are however, in this chapter, two significant pointers to the next.
The major part of the chapter considered the interaction as a small parameter
and expansions in powers of this small parameter were obtained. The van
der Waals treatment in Section 3.4 was an exception. That had no small
parameter and we noted that the resultant equation of state indicated some
non-physical behaviour in Fig. 3.19. We shall see that is indicative of a phase
transition.
In this context the hard sphere gas also deserves mention. True, this is an
expansion in a small parameter – here most sensibly regarded as density. And
the knowledge of a (limited) number of the virial coefficients tells us how the
resultant behavior of the gas is altered from that of the ideal gas. However
in both the Carnahan and Starling procedure and in the Padé approach all
terms of the virial expansion were approximated/estimated and the infinite
3.7. BRIDGE TO THE NEXT CHAPTER 177

series was summed to give an equation of state in closed form. Then, to

the extent that the higher-order coefficients are correct, the summed series
should indicate the behaviour at all densities up to the radius of convergence.

3.7.1 Van der Waals model

The lowest temperature van der Waals isotherm in Fig. 3.19 exhibits un-
physical behaviour. We see that for some pressures there are three possible
volumes. Moreover there is a region of the isotherm where ∂p/∂V is positive.
That is saying that when the pressure is increased the volume increases as
well. This is impossible; such a system would be unstable.
These matters will be discussed extensively in the next chapter. However
at this stage we simply point out that the calculated equation of state applies
fundamentally to a homogeneous system. And for certain values of p, T and
V homogeneous state may be unstable, with the corresponding stable state
comprising a coexistence of phases: a pointer to the next chapter.

3.7.2 Hard sphere model

We plot the Bannerman et al. molecular dynamics data yet again in Fig. 3.23.
This time we plot Z = pV /N kT on a logarithmic scale in order to include
the high density points omitted from the previous plots. We observe that
Z appears to diverge in the vicinity of y = 0.741. This corresponds to√the
density of periodic close packing of spheres (hcp or bcc lattice) at y = π/3 2 ,
shown as the line pcp.
The line rcp corresponds to the densest random close packing of spheres,
at y = 2/π ≈ 0.638; so this is the greatest density possible for a gas – or an
amorphous solid. However it is believed that the maximum possible density
of a gas is y = 0.495 and the minimum possible density of a solid is y = 0.545.
In other words, the density range 0.495 < y < 0.545 should be a gas-solid
coexistence region. The signature would be a constant pressure in this region.
The molecular dynamics data do not support this; rather, they indicate a
metastable supercooled gas. Presumably the simulations are not run long
enough for the equilibrium state to emerge. However the implication is that
the metastable state becomes unstable at y = 0.522.
Now although a random assembly of spheres will pack to a density of
y = 2/π ≈ 0.638, a close-packed
√ lattice (fcc or hcp) will pack more closely,
to a density of y = π/3 2 ≈ 0.740. So it is understandable that there is a
pressure drop at the transition from amorphous to regular solid.
Finally we note that the lack of an attractive part of the inter-particle
interaction means that there there is no liquid phase in these models; there
178 CHAPTER 3. NON-IDEAL GASES

Figure 3.23: Molecular dynamics simulations of the hard sphere equation of

state plotted on a logarithmic curve

is no gas-liquid transition. An essential feature of a liquid is that it be self-

bound. For this there has to be an attractive interaction.
3.7. BRIDGE TO THE NEXT CHAPTER 179

Further notes
Microscopic estimation of the van der Waals parameters
General form for the a parameter
The van der Waals a parameter is related to the mean attractive energy per
particle hEi by
V 2 d hEi
a= , (3.7.1)
N dV
where Z
1N
hEi = 4πr 2 U (r)dr (3.7.2)
2V
(from Eqs. (3.4.11) and (3.4.12)). Since the volume dependence is all in the
1/V prefactor, we obtain
Z
a = −2π r2 U (r)dr. (3.7.3)

For any two-parameter interaction energy function, with a scaling energy

ε and a scaling distance σ we have

U (r) = εu(r/σ), (3.7.4)

so that, by changing the variable of integration to x = r/σ,

Z
a = −εσ × 2π x2 u(x)dx.
3
(3.7.5)

The integral here is a dimensionless number, so we obtain the general result

a = εσ 3 × const. (3.7.6)

We conclude that the van der Waals a parameter will be given by εσ 3 mul-
tiplied by a model-dependent numerical factor.

Model-dependent prefactors
In Fig. 3.24 we show the Lennard-Jones 6 − 12 potential together with the
Sutherland potential. The Lennard-Jones potential has the familiar form, as
in Eq. (3.2.17)   
σ 12  σ 6
ULJ (r) = 4ε − .
r r
180 CHAPTER 3. NON-IDEAL GASES

Figure 3.24: Lennard-Jones (blue) and Sutherland (red) potentials. The

Sutherland potential has been scaled to that it coincides with the Lennard-
Jones potential at large distances.

However we now write the Sutherland potential in a form slightly different

to that in Eq. (3.2.25); here we express it as
US (r) = ∞ r<σ
 σ 6 (3.7.7)
= −4ε r > σ.
r
The factor of 4 here ensures that the Sutherland and the Lennard-Jones
potentials coincide at large differences.
Recall that the van der Waals σ parameter represents the effects of the
weak attractive part of the inter-particle interaction. So the range of the
integral in Eq. (3.7.5) might extend from r = σ up to infinity. In this case
16π 3
a= εσ = 5.59 × εσ 3 . (3.7.8)
9
Alternatively, we might go only from the minimum in the Lennard-Jones
potential, at r = rmin = 21/6 σ, up to infinity. In that case
√
10 2 π 3
a= εσ = 4.94 × εσ 3 . (3.7.9)
9
Since we are considering only the attractive part of the interaction it might
be more appropriate to consider the Sutherland potential. Integrating the
Sutherland potential (in its Eq. (3.7.7) form) over the entire range σ ≤ r < ∞
gives
8π 3
a= εσ = 8.38 × εσ 3 . (3.7.10)
3
3.7. BRIDGE TO THE NEXT CHAPTER 181

This is clearly a serious over-estimate as the integral incorporates short dis-

tances where the attractive potential falls well below the Lennard-Jones min-
imum of −ε; it falls to −4ε. We could remedy this by integrating only from
rmin = 21/6 σ, which gives
√
4 2π 3
a= εσ = 5.92 × εσ 3 . (3.7.11)
3
But this is still an overestimate and we might be inclined to integrate from
r0 = 21/3 σ, where the Sutherland potential has the Lennard-Jones minimum
of −ε. In that case
4π 3
a= εσ = 4.19 × εσ 3 . (3.7.12)
3
A different approach would be to look at the behaviour of the second
virial coefficient. The van der Waals equation of state leads to B2 as given
by Eq. (3.4.17):
a
B2VW (T ) = b − .
kT
By comparison, the attractive tail of the Lennard-Jones potential, as repre-
sented by the similar tail of the Sutherland Potential Eq. (3.7.7) will give a
second virial coefficient, at high temperatures, of
 
2 3 4ε
B2 (T ) = πσ 1 − . (3.7.13)
3 kT

Consistency of these two equations is ensured through the identification

8π 3
a= εσ = 8.38 × εσ 3 . (3.7.14)
3
And we note that this corresponds precisely to the value obtained from
integrating the Sutherland potential directly over the range σ ≤ r < ∞,
Eq. (3.7.10).

3.7.3 Other bit – Tabulation of - - - - -

Some values from Duda [22].

kTc /ε kTB /ε kTi /ε zc a

Square Well (R = 1.65) a a a a a
Lennard-Jones 1.326 3.418 6.431 0.281 a
Sutherland 0.595 1.171 2.252 0.303 a
182 CHAPTER 3. NON-IDEAL GASES

Problems for Chapter 3

3.1 Show that the Joule-Kelvin coefficient is zero for an ideal gas.

3.2 Derive the second

n virial coefficient
o expression for the Joule-Kelvin coef-
2T dB2 (T ) B2 (T )
ficient μJ = 5k dT
− T and find the inversion temperature for
the square well potential gas.

3.3 For the van der Waals gas show that TB = a/bk and Ti = 2a/bk.

3.4 Evaluate the constants A, B of Eq. (3.2.15) in terms of the square well
potential parameters σ and R.

3.5 Show that the Boyle temperature for the square well gas is given by

kTB = ε/ ln(1 − R−3 ).

3.6 Show that for the Lennard-Jones gas

TB = 3.42 ε/k, Ti = 6.43 ε/k

3.7 In Section 3.1.1 we saw that the partition function for an interacting gas
may be expressed as Z = Zid Q where Zid is the partition function for a
non-interacting gas and Q is the configuration integral. Explain why the
partition function of a hard sphere gas might be approximated by
 N
V − Nb
Z = Zid .
V

3.8 (a) Show that the approximate partition function for the hard sphere
gas in the previous question leads to the equation of state p (V − N b) =
N kT . This is sometimes called the Clausius equation of state. Give
a physical interpretation of this equation.
(b) Show that the first few virial coefficients are given by B2 (T ) =
b, B3 (T ) = b2 , B4 (T ) = b3 , etc. These virial coefficients are in-
dependent of temperature. Discuss whether this is a fundamental
property of the hard sphere gas, or whether it is simply a conse-
quence of the approximated partition function.
 3.7. BRIDGE TO THE NEXT CHAPTER 183

3.9 For a general interatomic interaction potential U (r ) we may define an

effective hard core dimension deff by U (deff ) = kT − Umin where Umin is
the minimum value of U (r). What is the significance of this definition?
Show that for the Lennard-Jones potential of Section 3.2.4, deff is given
by
( )1/6
2
deff = σ p , (3.7.15)
1 + 1 + kT /ε
corresponding to an effective volume
( )1/2
2
veff ∼ σ 3 p . (3.7.16)
1+ 1 + kT /ε

Show that for high temperatures veff ∼ (ε/kT )1/4 and compare this with
the high temperature limit of the Lennard-Jones B2 . Discuss the simi-
larities.
3.10 The one dimensional analogue of the hard sphere gas is an assembly of
rods constrained to move along a line (the Tonks model). For such a
gas of N rods of length l confined to a line of length L, evaluate the
configuration integral Q. Show that in the thermodynamic limit the
equation of state is
f (L − N l) = N kT (3.7.17)
where f is the force, the one dimensional analogue of pressure.
Comment on the similarities and the differences from the hard sphere
equation of state mentioned in Problem 3.8 (Clausius equation) and the
van der Waals equation of state.
3.11 Compare the square well and the van der Waals expressions for the second
virial coefficient. Show that they become equivalent when the range of
the square well potential tends to infinity while its depth tends to zero.
Show that R3 ε of the square well potential corresponds to 2ε of the van
der Waals “approximation” of the Lennard-Jones potential.
3.12 Show that the leading term in the expansion of g(n), of Eq. (3.6.8), is in
n2 i.e. show that there is no linear term.
3.13 From the expression for the second virial coefficient, Eq. (3.2.2), show
that if the interaction potential has a universal form U (r) = εu(r/σ)
— in other words u(x) is the same for different species, where just the
scaling parameters ε and σ differ — then B2 (T ) also has a universal form.
What are the relevant scaling parameters?
 184 CHAPTER 3. NON-IDEAL GASES
P
3.14 Evaluate the sum ∞ n
n=1 n(n + 3)y required in the derivation of the Car-
nahan and Starling hard sphere equation of state, Eq. (3.6.18). Hint:
P you
know the sum of the convergent geometric progression S(y) = ∞ n
n=0 y .
The trick of differentiation with respect to y will “bring down” an n into
the sum.

3.15 Eq. (3.6.26) gives the 3-2 Padé approximation to the hard sphere equation
of state based on knowledge of the first six virial coefficients. Expand
this expression in powers of y.

(a) Show that the expansion is consistent with the first six reduced
virial coefficients (Table 3.2).
(b) Compare the higher order terms with those given from the Carnahan
and Starling equation of state and compare both with the values
corresponding to the Clisby and McCoy virial coefficients.

Comment on whether (in your opinion) the 3-2 Padé or the Carnahan
and Starling is the better equation of state.

3.16 The interatomic potential U (r) = ε( σr )n may be regarded as a “soft

sphere” interaction where 1/n measures the “softness”.

(a) Show that when n → ∞ this reduces to the hard sphere potential.
(b) Plot U (r) for different n showing the hardening of the potential as
n increases.
(c) The second virial coefficient for particles interacting with this po-
tential is  
2 3  ε 3/n 3
B2 (T ) = πσ Γ 1− .
3 kT n
Show this reduces to the hard sphere case when n → ∞.
(d) Sketch the form of this B2 (T ) together with the general form for
the second virial coefficient. They are very different. Explain why.
(e) For the hard sphere gas we saw that the compressibility factor
pV /N kT could be expressed as a universal, temperature-independent,
function of the density N/V . By noting that, for a given value of
n, the interaction potential scales with a single parameter εσ n , ar-
gue that the compressibility factor z may now be expressed as a
universal function of N/V T 3/n

3.17 Some measurements of the second virial coefficient B2 (T ) for xenon at

high temperatures are given in Table 3.4.
3.7. BRIDGE TO THE NEXT CHAPTER 185

Table 3.4: Second virial coefficient of xenon

T (K) B2 (cm3 /mol)

273.16 −154.74
298.16 −130.21
323.16 −110.62
348.16 −95.04
373.16 −82.13
423.16 −62.10
473.16 −46.74
573.16 −25.06
673.16 −10.77
773.16 −0.13
873.16 7.95
973.16 14.22

If you assume that the interaction between two xenon atoms has the form
of the square well potential, what can you deduce about the potential’s
parameters σ, ε and R?
Answer: hghjgjggggj
At high temperatures
 
2 3 3 ε (R3 − 1)  ε 2 (R3 − 1)  ε 3
B2 (T ) = πσ 1 − (R − 1) − − + ...
3 kT 2 kT 6 kT
(3.7.18)
186 CHAPTER 3. NON-IDEAL GASES

Figure 3.25: Fit of square well expression through xenon data points.

know how to do the references. ”ieeetr” numbers the references in order

of appearance (”plain” numbers alphabetically)

Bibliography
[1] W. J. Mullin, “The cluster expansion,” Am. J. Phys., vol. 40, p. 1473,
1972.

[2] L. E. Reichl, A Modern Course in Statistical Mechanics. John Wiley,

1998.

[3] I. Cachadiña Gutiérrez, “Comment on ‘Exact Analytic Second

Virial Coefficient for the Lennard-Jones Fluid’ (arXiv:0909.3326v1).”
arXiv:1310.3586v1, Oct. 2013.

[4] C. Kittel, Introduction to Solid State Physics. John Wiley, New York,
7th. ed., 1996.

[5] W. J. Mullin and G. Blaylock, “Quantum statistics: is there an effective

fermion repulsion or boson attraction?,” Am. J. Phys., vol. 71, p. 1223,
2003.

[6] T. D. Lee and C. N. Yang, “Many-body problem in quantum statistical

mechanics. I. General formulation,” Physical Review, 1959.
BIBLIOGRAPHY 187

[7] S. Y. Larsen, J. E. Kilpatrick, E. H. Lieb, and H. F. Jordan, “Sup-

pression at High Temperature of Effects Due to Statistics in the Second
Virial Coefficient of a Real Gas,” Physical Review, 1965.

[8] R. P. Feynman, Statistical Mechanics. Benjamin, 1972.

[9] R. Young, “Theory of quantum-mechanical effects on the thermody-

namic properties of Lennard-Jones fluids,” Physical Review A, vol. 23,
pp. 1498–1510, Jan. 1981.

[10] M. W. Zemansky, Heat and Thermodynamics. McGraw-Hill, 1968.

[11] J. S. Rowlinson, J. D. van der Waals: On the Continuity of the Gaseous

and the Liquid States, Studies in Statistical Mechanics XIV. North Hol-
land, 1988.

[12] L. D. Landau and E. M. Lifshitz, Statistical Physics, Part 2. Pergamon

Press, 1980.

[13] H. J. Wintle, “More on the Equation of State for Gases,” Am. J. Phys,
vol. 42, no. 9, p. 794, 1974.

[14] P. M. Chaikin and T. C. Lubensky, Principles of Condensed Matter

Physics. Cambridge University Press, 1995.

[15] F. Reif, Fundamentals of Statistical and Thermal Physics. McGraw-Hill,

1965.

[16] M. N. Bannerman, L. Lue, and L. V. Woodcock, “Thermodynamic pres-

sures for hard spheres and closed-virial equation-of-state,” J. Chem.
Phys., vol. 132, p. 084507, 2010.

[17] H. Gould and J. Tobochnik, “Hard disc simulation.”

http://stp.clarku.edu/simulations/harddisks/dynamics/index.html.

[18] F. H. Ree and W. G. Hoover, “Fifth and sixth virial coefficients for hard
spheres and hard disks,” J. Chem. Phys., vol. 40, p. 939, 1964.

[19] N. Clisby and B. McCoy, “Ninth and tenth order virial coefficients for
hard spheres in D dimensions,” J. Stat. Phys., vol. 122, p. 15, 2006.

[20] R. J. Wheatley, “Calculation of high-order virial coefficients with ap-

plications to hard and soft spheres,” Physical review letters, vol. 110,
no. 20, p. 200601, 2013.
188 CHAPTER 3. NON-IDEAL GASES

[21] N. F. Carnahan and K. E. Starling, “Equation of state for non-attracting

rigid spheres,” J. Chem. Phys., vol. 51, p. 635, 1969.

[22] Y. Duda and P. Orea, “New criteria for the equation of state develop-
ment: Simple model fluids,” Fluid Phase Equilibria, vol. 272, pp. 93–95,
Oct. 2008.