You are on page 1of 8

1

CHEM-UA 652: Thermodynamics and Kinetics


Notes for Lecture 3
I. GENERALIZING THE EQUATION OF STATE
The van der Waals equation of state corresponds to one particular model for the interparticle forces in a real gas. Is
it possible to come up with a more general type of equation of state that could be t to data either from experiments
or very accurate computer simulations? It turns out that there is, and the van der Waals equation gives a clue as to
how such an equation of state might appear.
Consider the compressibility of the van der Waals equation:
P
RT
= Z =
1
1 b

a
RT
(1)
If the density is small, then the rst term can be expanded in a power series in the density, using the fact that
1
1 x
=

k=0
x
k
(2)
from which
P
RT
=

k=0
b
k

a
RT
= 1 +
_
b
a
RT
_
+

k=2
b
k

k
(3)
In this expression, the coecient of the term linear in is a function of the temperature T, however, for the remaining
powers of , i.e.,
2
,
3
,..., the coecients are just b
k
. This is a consequence of the approximations inherent in the
van der Waals equation model, which we have seen leads to some serious physical inconsistencies.
More realistic treatments of non-ideal gases generate an equation of state in the form of a power series in the density
in which all of the coecients are functions of temperature. This can be expressed as
P
RT
= 1 +

k=1
B
k+1
(T)
k
(4)
This equation of state is called the virial equation of state. The coecients B
k+1
(T) are called virial coecients.
In the low density limit, the most important term in this power series is the k = 1 or linear term, which gives the
approximate equation of state
P
RT
1 +B
2
(T) (5)
where the dominant coecient B
2
(T) is called the second virial coecient. For the van der Waals equation, we see
that B
2
(T) = b (a/RT). As a function of T, this looks as follows: The following plot shows the second virial
coecient for several gases. Although it is dicult to see, these curves pass through a shallow maximum that is not
captured by the van der Waals equation. Thus, the van der Waals equation is only in qualitative agreement with
these but is not fully quantitative.
In order to do better, we need a better model for the interaction potential u(r). If we have a reliable representation
for u(r), then in the low density limit, the second virial coecient can be obtained approximately from the formula
B
2
(T) 2N
0
_

0
_
e
u(r)
1
_
r
2
dr (6)
2
T
B
2
(
T
)
b
FIG. 1: Second virial coecient for the van der Waals equation.
where = 1/k
B
T.
Let us look at some simple examples of B
2
(T) calculations.
1. Hard-sphere potential:
u(r) =
_
r
0 r >
(7)
For this potential, the calculation of B
2
(T) proceeds as follows:
B
2
(T) = 2N
0
__

0
_
e

1
_
r
2
dr +
_

_
e
0
1
_
r
2
dr
_
= 2N
0
__

0
(0 1) r
2
dr +
_

(1 1) r
2
dr
_
= 2N
0
_

0
r
2
dr
=
2
3

3
N
0
(8)
3
FIG. 2: Experimental second virial coecient plots for several real gases.
which is also the paramter b in the van der Waals equation. In this case, there is no temperature dependence in
the result for B
2
(T).
2. Square-well potential plus hard-sphere potential: In this case, the potential takes the form
u(r) =
_
r
r
0 r >
(9)
In this case, the hard-sphere potential is supplemented by a short-range constant attractive potential that has
a square-well shape. The parameter > 1 determins the range of the square well. For this example, the
calculation of B
2
(T) proceeds as follows:
B
2
(T) = 2N
0
_
_

0
_
e

1
_
r
2
dr +
_

_
e

1
_
r
2
dr +
_

_
e
0
1
_
r
2
dr
_
= 2N
0
_

_

0
r
2
dr +
_
e

1
_
_

r
2
dr
_
= 2N
0
_

3
3
+
_
e

1
_
r
3
3

_
= 2N
0
_

3
3
+
_
e

1
_
1
3
_

3
_
_
=
2N
0

3
3
_
1
_
e

1
_ _

3
1
_
(10)
A plot of this second virial coecient is given in the gure below.
4
0
2
3
N
0
/3
FIG. 3: Second virial coecient for the hard-sphere plus square-well potential.
3. Hard-sphere plus C
6
/r
6
potential: For this example, the potential is
u(r) =
_
r

C6
r
6
r >
(11)
This is a particular realization of a potential that leads to the van der Waals equation of state. The C
6
/r
6
attractive part of the potential comes directly from a full quantum mechanical treatment of the electron dis-
tribution around each atom in an interacting pair. Such an attractive potential attempts to model London
dispersion or van der Waals forces. For this example, the calculation of B
2
(T) proceeds as follows:
B
2
(T) = 2N
0
__

0
_
e

1
_
r
2
dr +
_

_
e
C6/r
2
1
_
r
2
dr
_
= 2N
0
_

3
3
+
_

_
e
C6/r
2
1
_
r
2
dr
_
(12)
he remaining integral cannot be evaluated in closed form, but let us suppose that C
6
/r
6
<< 1 for all values of r
considered, i.e., r < . Then, we can expand the exponential using the power series e
x
1+x+x
2
/2! + .
If we retain just the rst two terms, then we have the approximation
e
C6/r
6
1 1 +
C
6
r
6
1 =
C
6
r
6
(13)
5
With this approximation, the calculation of B
2
(T) can proceed as follows:
B
2
(T) 2N
0
_

3
3
+C
6
_

1
r
6
r
2
dr
_
= 2N
0
_

3
3
+C
6
_

1
r
4
dr
_
= 2N
0
_

3
3
+C
6
_

1
3r
3

__
= 2N
0
_

3
3
+
C
6
3
3
_
=
2N
0

3
3
_
1
C
6
3k
B
T
6
_
(14)
A plot of this function will follow the curve presented in Fig. 1.
However, in this example, we can improve the approximation by retaining more terms of the exponential we
expanded above. In fact, let us see what happens if we retain all of the terms in the power series. That is, we
will use the fact that e
x
can be represented exactly as
e
x
=

k=0
x
k
k!
= 1 +

k=1
x
k
k!
(15)
so that
e
C6/r
6
1 =

k=1
1
k!
_
C
6
r
6
_
k
= 1 +

k=1

k
C
k
6
k!
1
r
6k
(16)
With the expansion, the calculation of B
2
(T) proceeds as follows:
B
2
(T) = 2N
0
_

3
3
+

k=1

k
C
k
6
k!
_

1
r
6k
r
2
dr
_
= 2N
0
_

3
3
+

k=1

k
C
k
6
k!
_

1
r
6k2
dr
_
= 2N
0
_

3
3
+

k=1

k
C
k
6
k!
1
(6k 1)
6k1
_
=
2N
0

3
3
_
1 3

k=1
(C
6
)
k
k!
1
(6k 1)
6k+2
_
(17)
The nal result represents and exact expression for B
2
(T) for this particular model of u(r). How does it improve
on the result obtained above retaining just two terms in the expansion of the exponential? Let us plot the nal
result as a function of temperature in the following way: Unfortunately, we cannot resum the nal expression
into a nice closed form, but we can evaluate the sum numerically on a computer. However, on a computer, we
cannot sum an innite number of terms. Rather, we replace the upper limit of with a maximum number of
terms K
max
and then just increase K
max
until the result converges. So, lets do that for several values of K
max
and see how the curve changes as we increase K
max
. This will show us how the result converges with K
max
.
The result is shown in the gure below. We see from the gure that the sum converges very rapidly with K
max
6
0
T
0
B
2
(
T
)
K
max
= 2
K
max
= 3
K
max
= 4
K
max
= 5
K
max
= 6
2N
0

3
/3
FIG. 4: Second virial coecient for the potential in Eq.11.
and only requres a few terms. Not unexpectedly, the most signicant deviations from K
max
= 2 occur in the
low-temperature part of the curves.
4. Perhaps the most realistic representation of the potential u(r) shown in Fig. 2 of Lecture 2 is the so-called
Lennard-Jones potential, for which
u(r) = 4
_
_

r
_
12

r
_
6
_
(18)
A plot of this potential very closely resembles that shown in Fig. 2 of Lecture 2. In particular, there is a
well-dened minimum, which we can compute by taking the derivative of u(r), setting it to 0, and solving for r
at that point:
U

(r) = 4
_
12
12
r
13

6
6
r
7
_
= 0
12
12
r
13
=
6
6
r
7
2
6
r
13
=
1
r
7
7
2
6
= r
6
r = 2
1/6
r
m
(19)
where r
m
denotes the location of the minimum. The value of the Lennard-Jones potential at that point is
u(r
m
) = u(2
1/6
) = 4
_
_

2
1/6

_
12
_

2
1/6

_
6
_
= 4
_
1
2
2

1
2
_
= 4
1
4
= (20)
Thus, the parameter measures the depth of the minimum at r = r
m
. In addition, u() = 0, and for r < , the
potential exhibits a very steep increase, very much like the hard-sphere potential. Hence, is a measure of the
van der Waals radius of each particle.
The calculation of the second virial coecient can now be set up as
B
2
(T) = 2N
0
_

0
_
exp
_
4
_
_

r
_
12

r
_
6
_
1
__
r
2
dr (21)
Let us make a change of variables from r to
x =
r

so that dr = dx. Let us also dene

= . With this change, the integral for B


2
(T) becomes
B
2
(T) = 2
3
N
0
_

0
_
e
4

(x
12
x
6
)
1
_
x
2
dx (22)
Again, this integral cannot be evaluated analytically in closed form. We could expand the exponential in an
innite series as we did in the previous example. However, if we were to do this, the evaluation of the various
terms in the integral would be considerably more complicated than in the previous example, as we would
need to evaluated increasingly higher powers of a binomial form (x
12
x
6
)
k
, Instead, since this is just a
one-dimensional integral, we can evaluate it numerically using a method such as Simpsons rule. Recall that
Simpsons rule for the integral of a function f(x) using a set of n evenly spaced values for x (x
0
, x
1
,...,x
n
) is
just
_
b
a
f(x)dx =
h
3
[f(x
0
) + 4f(x
1
) + 2f(x
2
) + 4f(x
3
) + 2f(x
4
) + 2f(x
n2
) + 4f(x
n1
) +f(x
n
)] (23)
where h = (b a)/n, and the points x
0
,...,x
n
are given by x
j
= a +jh, j = 0, ..., n.
Applying this to the integral in Eq. (22), the resulting curve for B
2
(T) as a function of T appears as in the
gure below: In this gure, k
B
T

= 1/

, and
B

2
(T) =
B
2
(T)
2
3
N
0
/3
Unfortunately, the curve in the book does not carry the curve out to high enough temperature to see the
maximum. However, on page 233 of the book Statistical Mechanics by Donald A. McQuarrie, a more complete
curve for the Lennard-Jones potential is shown, and in this curve the temperature axis is suciently long to see
the maximum. This book is available on Google Books.
8
FIG. 5: Second virial coecient for the Lennard-Jones potential evaluated using numerical integration. Experimental points
are also plotted. Because of the use of reduced units, data for all gases land on a single master curve, as an example of the
law of corresponding states.

You might also like