CHEM0019: Statistical Mechanics

Prof Katherine B. Holt

[email protected]

Overview of course:

• Topic 1: Energy states and populations.

• Topic 2: Molecular partition functions.

• Topic 3: Towards thermodynamic functions.

• Topic 4: Translational partition function.

• Topic 5: Rotational partition function and spin statistics.

• Topic 6: Other partition functions.

• Topic 7: Weights and configurations.

• Topic 8: Statistical entropy.

Recommended Texts (All Available as e-books via BibliU):

‘Statistical Thermodynamics’ A. Maczek, Oxford Chemistry Primer.

‘Atkins’ Physical Chemistry’ P. W. Atkins, OUP, Chapters 13A – 13F.

1
Topic 1: Energy states and populations

1.1: What is Statistical Mechanics?

Often called Statistical Thermodynamics, this topic is a bridge between the

microscopic properties of matter considered in quantum mechanics and
spectroscopy and the bulk properties encountered in the study of
thermodynamics. By the end of this course you will be able to calculate
thermodynamic properties, such as internal energy and Gibb’s energy, using
spectroscopic data and equations derived during the quantum mechanics
course. How incredible is that! You will also have an insight into how bulk
chemical properties like Gibb’s energy have a molecular origin.

1.2. The Boltzmann Distribution

The Boltzmann distribution is one of the most important equations in chemistry

(maybe the most important – what do you think?). It gives the number of
molecules (𝑛𝑖 ) in a particular energy state (𝜀𝑖 ) at temperature T:
𝜀𝑖
𝑛𝑖 ∝ e− 𝑘𝑇 𝐸𝑞. 1.1

𝑘 in this equation is Boltzmann’s constant = 1.3806488 x 10-23 J K-1.

The Figure below shows the distribution of molecules across five states of
increasing energy as temperature is increased from T = 0 K to T = ∞.

2
At T = 0 K all molecules occupy the lowest energy state but as temperature is
increased more molecules have enough energy to populate higher states. As
T tends to infinity all energy levels become equally populated. [Note: a
common error is to imagine that as T → ∞ all the molecules will be found in
the upper energy state; this does not happen – all states become equally
populated.]
To remove the need for a proportionality constant in Eq. 1.1 we often consider
the relative populations of the states, or the ratio of the number of molecules in
state 𝑖 (𝑛𝑖 ) with energy 𝜀𝑖 to the number in state 𝑗 (𝑛𝑗 ) with energy 𝜀𝑗 (Eq. 1.2):
(𝜀𝑖 −𝜀𝑗 )
𝑛𝑖 −
= e 𝑘𝑇 𝐸𝑞. 1.2
𝑛𝑗
An understanding of how energy is distributed according to the Boltzmann
distribution is key to understanding most aspects of thermodynamics and
kinetics in chemistry. It is clear that temperature is the only parameter that
governs the population of the available energy states.

1.3. The partition function

From Eq. 1.2 the relative number of molecules occupying a particular energy
state can be determined; however, it is not possible to obtain exact values for
𝑛𝑖 and 𝑛𝑗 . A value we do often know is the total number of molecules 𝑁. (For
example, this could be Avogadro’s number (𝑁𝐴 or 𝐿) for 1 mol of molecules.) 𝑁
is equal to the sum of the number of molecules in each energy level:

𝑁 = 𝑛0 + 𝑛1 + 𝑛2 + ⋯ = ∑ 𝑛𝑖 𝐸𝑞. 1.3.
𝑖

By convention we label the lowest energy state 𝜀0 (with number of molecules

in that state being 𝑛0 ). By convention we also assume that 𝜀0 = 0.
Rewriting Eq 1.2 assuming that 𝜀𝑗 is the lowest energy state 𝜀0 and that 𝜀0 = 0:
1
𝑛𝑖 = 𝑛0 e−𝛽𝜀𝑖 where 𝛽 = 𝐸𝑞. 1.4
𝑘𝑇
Substituting Eq. 1.4 into Eq. 1.3 gives:

𝑁 = ∑ 𝑛0 e−𝛽𝜀𝑖 = 𝑛0 ∑ e−𝛽𝜀𝑖
𝑖 𝑖

3
And rearranging for 𝑛0 gives:
𝑁
𝑛0 = Eq. 1.5
∑𝑖 e−𝛽𝜀𝑖

Eq. 1.5 can now be substituted into Eq. 1.4:

𝑁e−𝛽𝜀𝑖
𝑛𝑖 = Eq. 1.6
∑𝑖 e−𝛽𝜀𝑖

And by employing the definition:

𝑞 = ∑ e−𝛽𝜀𝑖 𝐸𝑞. 1.7

𝑖

Allows us to also write Eq 1.6 as:

𝑁e−𝛽𝜀𝑖
𝑛𝑖 = 𝐸𝑞. 1.6. 𝐴
𝑞

Where 𝑞 is called the partition function.

So, what is the partition function? What does it represent? At present it is

merely a convenient abbreviation for the sum over all energy states of e−𝛽𝜀𝑖 .
However, as we continue through this subject, we will discover that it
measures how the total number of molecules is distributed (or partitioned) over
the available states. It contains all the information needed to calculate the bulk
properties of a system composed of independent molecules.

Now look at worked problems 1 (WP 1.1-1.3) where some examples of how to
obtain and interpret the partition function are discussed.

4
Topic 2: Molecular partition functions

2.1. Molecular energy levels

In Topic 1 the energy states were treated in a very generic way and the form
or type of energy was not defined. This is because one of the principles of
Statistical Mechanics is the very powerful ‘principle of equal a priori
probabilities’. This is a complex way of stating the simple assumption that all
possibilities for the distribution of energy are equally probable. Thus, the
treatment of energy states in Topic 1 could be used for any types of molecular
energy – e.g. two conformations with different energy, different spin states,
different electronic energies etc.
In Topic 2 we are going to focus on the population of energy states concerned
with molecular motion: these are translational, rotational and vibrational
energy levels. The ‘principle of equal a priori probabilities’ allows us to
assume that if (for example) a rotation state and a vibrational state have the
same energy they have an equal probability of being occupied.
So first let’s revise what we know about the energy levels for molecular motion
from quantum mechanics and spectroscopy.
Vibrational energy levels can be derived by solving the Schrodinger
Equation for the Simple Harmonic Oscillator (SHO). This results in a series of
evenly spaced non-degenerate energy levels at 𝐸 = (𝑣 + 1⁄2)ℏ𝜔 (in J). Hence
the energy difference between levels is Δ𝐸 = ℏ𝜔.
In spectroscopy the vibrational energy levels are usually reported as a term
𝐺(𝑣) = 𝜔𝑒 (𝑣 + 1⁄2) determined in wavenumbers (cm-1). The spacing between
levels is therefore Δ𝐺(𝑣) = 𝜔𝑒 . As shown in spectroscopy lectures, typical
values of 𝜔𝑒 are 1000 – 3000 cm-1 (e.g. 𝜔𝑒 for HCl is 2987 cm-1).
Of course the SHO model does not take into account anharmonicity and the
fact that real bonds eventually break when vibrationally excited above the
dissociation energy. However, in the vibrational ground state the SHO model
is acceptable and hence we will use the resulting energy level expressions in
further discussions of the vibrational partition function 𝑞 𝑉 .
Rotational energy levels can be derived by solving the Schrodinger Equation
for a particle on a sphere or rigid rotor. This results in a series of energy levels
whose separation increases with increasing energy at 𝐸 = (ℏ⁄2𝐼) 𝐽(𝐽 + 1).

5
In spectroscopy the rotational levels are described in cm-1 as 𝐹(𝐽) = 𝐵𝐽(𝐽 + 1)
and difference between consecutive levels increases from 2B to 4B to 6B etc.
Typical values of B range from 0.1 to 60 cm-1 leading to differences between
lower rotational levels in the order of 10-1 to 102 cm-1.
Unlike vibrational energy levels, rotational energy levels exhibit multiple
degeneracy, with (2𝐽 + 1) states available at each energy level.
Translational energy levels can be derived and described using the particle
in a box model (see Topic 4) but for most practical purposes the energy levels
for a typical molecule in typical volumes at typical temperatures are so close
together as to be considered a continuum. As we experience in our everyday
lives, the translational motion of gases can be treated classically in most
situations. Exceptions would be very light molecules at low temperatures in
confined volumes.

2.2. Levels and states

You may or may not have noticed that I used the term ‘energy states’
throughout Topic 1, yet in section 2.1 the term ‘energy level’ was used to
describe specific molecular energies. It is important in statistical mechanics to
distinguish between the two. This is because a number of states (𝑔𝑖 ) may be
available at an energy (𝜀𝑖 ) and hence that energy level is 𝑔𝑖 -fold degenerate.
The Figure below shows an example where there are two energy levels, each
of which is 2-fold degenerate (i.e. 𝑔0 = 2 and 𝑔1 = 2).

If instead of determining the partition function over states we now sum over
levels we need to take into account the degeneracy and so we can write:

𝑞= ∑ 𝑔𝑖 e−𝛽𝜀𝑖 𝐸𝑞. 2.1 (𝑐. 𝑓. 𝐸𝑞. 1.7)

𝑖 (𝑙𝑒𝑣𝑒𝑙𝑠)

6
2.3. Contributions to the partition function

From this point onwards I am going to refer to 𝑞 as the molecular partition

function as we’ll specifically be considering how energy is partitioned among
molecular energy levels. Each of the different modes of molecular energy can
contribute to 𝑞.
The Born-Oppenheimer approximation, which you have encountered
previously in spectroscopy, allows us to sum the contributions from different
modes of energy to obtain the total energy of an isolated molecule. Hence:
𝜀𝑖 = 𝜀𝑖𝑇 + 𝜀𝑖𝑅 + 𝜀𝑖𝑉 𝐸𝑞. 2.2
Where 𝑇 denotes translation, 𝑅 rotation and 𝑉 vibration. The separation of the
vibrational and rotational terms is an approximation but in most cases is
satisfactory.

Hence, we can write an expression for the molecular partition function:

𝑇 𝑅 𝑉 𝑇 𝑅 𝑉
𝑞= ∑ e−𝛽𝜀𝑖 = ∑ e−𝛽(𝜀𝑖 +𝜀𝑖 +𝜀𝑖 ) = ∑ e−𝛽𝜀𝑖 ∑ e−𝛽𝜀𝑖 ∑ e−𝛽𝜀𝑖
𝑖 (𝑠𝑡𝑎𝑡𝑒𝑠) 𝑖 (𝑠𝑡𝑎𝑡𝑒𝑠) 𝑖(𝑇) 𝑖(𝑅) 𝑖(𝑉)

Clearly this shows:

𝑞 = 𝑞 𝑇 𝑞 𝑅 𝑞𝑉 𝐸𝑞. 2.3

This factorisation allows us to investigate each contribution separately and in

future Topics (and worked problems and tutorial questions) we will derive
some approximate expressions for 𝑞 𝑇 , 𝑞 𝑅 and 𝑞 𝑉 and calculate the values for
different molecules.

For now let’s consider typical values of 𝑞 𝑇 , 𝑞 𝑅 and 𝑞 𝑉 for 1H35Cl at room
temperature (298 K) so that we can appreciate how energy is distributed /
partitioned across different modes in typical diatomic molecule. The
calculations are explained in more detail in Worked Problems 2.

𝑉
𝑞 𝑉 = ∑ e−𝛽𝜀𝑖 = ∑ e−𝛽𝑣ℎ𝑐𝜔𝑒
𝑖 𝑣

And as 𝜔𝑒 = 2987 cm-1 𝑞 𝑉 = 1.0000005

7
 𝑅
𝑞 𝑅 = ∑ e−𝛽𝜀𝑖 = ∑(2𝐽 + 1)e−𝛽𝐵ℎ𝑐𝐽(𝐽+1)
𝑖 𝐽

And as B = 10.591 cm-1 𝑞 𝑅 = 19.9

𝑇 (𝑋) (𝑌) (𝑍)

𝑞 𝑇 = ∑ e−𝛽𝜀𝑖 = ∑ e−𝛽𝐸𝑛 ∑ e−𝛽𝐸𝑛 ∑ e−𝛽𝐸𝑛
𝑖 𝑛(𝑋) 𝑛(𝑌) 𝑛(𝑍)

𝑞 𝑇 ~ 1032 in a 1 m3 volume container.

So, what do these values tell us about how energy partitions in a typical
molecule? First, it is clear that 𝑞 𝑉 is very small (close to 1) and 𝑞 𝑇 is huge –
many, many orders of magnitude greater. Thus, many more translational
energy levels are occupied at room temperature compared to vibrational (and
rotational) states.

Why is this? As discussed above, translational energy levels are very close
together in energy and the separation between them is far smaller than the
thermal energy available to the molecules at room temperature.
∆𝜀 𝑇 ≪ 𝑘𝑇
Hence the molecules have sufficient energy to occupy many translational
states and 𝑞 𝑇 is consequentially large.

On the contrary, the energy difference between vibrational levels is large

compared to the thermal energy available at room temperature.
∆𝜀 𝑉 ≫ 𝑘𝑇
Hence most molecules will occupy the lowest energy level (vibrational ground
state) at 298 K. This is consistent with observations from spectroscopy.

So, the ‘take home message’ is:

The molecular partition function 𝑞 gives an indication of the number of states

that are thermally accessible to a molecule at the temperature of the system.

8
Topic 3: Towards thermodynamic functions

3.1. Mean molecular energy

For statistical mechanics to be useful, we need to use the partition function to

extract thermodynamic information. A simple place to start is to consider the
mean energy of a molecule 〈𝜀〉.

An important first assumption: we have a total of 𝑁 molecules that are non-

interacting (independent). Hence if each molecule occupies a state with
energy 𝜀𝑖 the total energy 𝐸 is:

𝐸 = ∑ 𝑛𝑖 𝜀𝑖 𝐸𝑞. 3.1
𝑖

Then the mean (average) energy of a molecule is:

𝐸 1
〈𝜀〉 = = ∑ 𝑛𝑖 𝜀𝑖 𝐸𝑞. 3.2
𝑁 𝑁
𝑖

Looking back to Topic 1 we already encountered an expression for 𝑛𝑖 :

𝑁e−𝛽𝜀𝑖
𝑛𝑖 = 𝐸𝑞. 1.6. 𝐴
𝑞
That we can now substitute into Eq. 3.2:

1
〈𝜀〉 = ∑ 𝜀𝑖 e−𝛽𝜀𝑖 𝐸𝑞. 3.3
𝑞
𝑖
To manipulate this equation into a form that involves only 𝑞 (i.e. that eliminates
𝜀𝑖 ) we can note the following relationship:
d −𝛽𝜀
𝜀𝑖 e−𝛽𝜀𝑖 = − e 𝑖
d𝛽
Hence it follows that:
1 d −𝛽𝜀 1 d 1 d𝑞
〈𝜀〉 = − ∑ e 𝑖=− ∑ e−𝛽𝜀𝑖 = − 𝐸𝑞. 3.4
𝑞 d𝛽 𝑞 d𝛽 𝑞 d𝛽
𝑖 𝑖
So now you can see that we have derived an expression for mean molecular
energy only in terms of the partition function (and its temperature dependence
– remember 𝛽 = 1⁄𝑘𝑇).

9
However, we need to make two modifications to Eq 3.4. First, by convention
we have assumed that 𝜀0 = 0 when determining 𝑞. This is not always the case
if 𝜀𝑔𝑠 ≠ 0 (for example for vibration 𝜀𝑔𝑠 is the zero point energy for the SHO).
Hence the true mean energy would be 〈𝜀〉 + 𝜀𝑔𝑠 . Secondly, in some cases the
partition function might depend on the volume as well as temperature (this is
true for 𝑞 𝑇 ) so we need to re-write the derivative in Eq 3.4 as a partial
derivative. Thus, the complete expression relating the partition function to the
mean energy of a molecule is:
1 𝜕𝑞
〈𝜀〉 = 𝜀𝑔𝑠 − ( ) 𝐸𝑞. 3.5
𝑞 𝜕𝛽 𝑉
A demonstration of how to use this expression is given in Worked Problems 3.

3.2. Internal energy of non-interacting molecules

If we continue with the assumption that molecules are non-interacting, we can
assume that total energy for 𝑁 molecules (𝐸𝑁 ) is given by:

𝐸𝑁 = 𝑁〈𝜀〉 𝐸𝑞. 3.6

And by noting that 𝐸𝑁 = 𝑈 where 𝑈 is the internal energy:

𝑁 𝜕𝑞
𝑈 = 𝑈(0) − ( ) 𝐸𝑞. 3.7
𝑞 𝜕𝛽 𝑉
This can also be written as:
𝜕ln𝑞
𝑈 = 𝑈(0) − 𝑁 ( ) 𝐸𝑞. 3.7𝐴
𝜕𝛽 𝑉
And
𝜕ln𝑞
𝑈 = 𝑈(0) + 𝑁𝑘𝑇 2 ( ) 𝐸𝑞. 3.7𝐵
𝜕𝑇 𝑉
The derivation of Eq 3.7 A and 3.7 B are discussed further in the Worked
Problems for this Topic.

From the expressions for 𝑈, other equations for thermodynamic functions

(𝐶𝑉 , 𝑆, 𝐻, 𝐺 etc) can be derived in terms of 𝑞. We don’t have time to derive
these expressions in this course so some of these are listed in section 3.5 and
we will apply these in later Topics.

10
3.3. The canonical ensemble and canonical partition function

We made the assumption above that the molecules were non-interacting but in
practice molecules do interact (real gases, liquids, solids etc) and we need to
construct a model to account for this. In Eq. 3.6 we assumed we can simply
scale the total energy of the system by multiplying the average energy of each
molecule by the number of molecules there are in the system. However, if
molecules interact with each other (for example electrostatic attractions and
repulsions, hydrogen-bonding etc) additional energy terms arise that do not
scale linearly with the number of molecules present.
The way around this is to imagine a closed collection of 𝑁 interacting
molecules at a specific volume and temperature. We imagine additionally that
these molecules can be distributed across a set of energy states and that each
molecule has a total energy 𝐸𝑖 . These energy states can adjust to any
intermolecular interactions that exist. We then replicate this system multiple
times and consider the behaviour of this collection of replications. This
imaginary collection is called the canonical ensemble and allows us to define
this partitioning of the molecules among the available system energy states as
the canonical partition function 𝑄.
In the same way that 𝑞 allows the calculation of mean molecular energy, so 𝑄
can be used to calculate the mean energy of an entire system composed of
molecules that may or may not be interacting with each other. Thus:

𝑄 = ∑ e−𝛽𝐸𝑖 𝐸𝑞. 3.8

𝑖

Where the sum here is over all members of the ensemble. 𝑄 is more versatile
than 𝑞 because molecular interactions are already ‘baked in’ to the definition
and so it can be applied to condensed phases and real gases. Thus, the
energy of an ensemble of interacting molecules is given by:
𝜕ln𝑄
𝑈 = 𝑈(0) − ( ) 𝐸𝑞. 3.9
𝜕𝛽 𝑉

Note that by comparison with Eq 3.7 A the scaling factor of 𝑁 is not required in
Eq 3.9, as 𝑄 already considers the partitioning of the 𝑁 molecules as a whole
and not as the product of 𝑁 separate molecular contributions.

11
3.4. Relationship between canonical and molecular partition functions

We are now (for simplicity) going to return to considering non-interacting

molecules and will continue to do so for the rest of the Statistical Mechanics
course. When molecules are non-interacting or independent 𝑄 can be shown
to have relatively simple relationships to 𝑞.

When molecules are independent and distinguishable:

𝑄 = 𝑞𝑁 𝐸𝑞. 3.10

When molecules are independent and indistinguishable:

𝑞𝑁
𝑄= 𝐸𝑞. 3.11
𝑁!
So, what are distinguishable and indistinguishable molecules? In a solid,
molecules are distinguishable even when identical because they have a
defined position in a crystal lattice and can be ‘labelled’ or ‘named’ with a set
of coordinates. In that case Eq 3.10 can be used. However, in a gas or liquid
where molecules are in constant motion, they become indistinguishable and
Eq 3.11 must be used. We will not go further into the origins of Eqs 3.10 and
3.11 but you can read more if interesting in both Recommended texts. Worked
problem 3.3 touches on why Eq. 3.11 should be used for indistinguishable
molecules to prevent over-counting.
It turns out that when determining internal energy, it doesn’t matter which of
Eq 3.10 or 3.11 you use, as substituting Eq 3.11 into 3.9 gives:

𝑞𝑁
𝜕ln ( ) 𝜕(ln𝑞𝑁 − ln𝑁!) 𝜕(𝑁ln𝑞) 𝜕ln𝑞
𝑁!
( )
𝑈 − 𝑈 0 = −( ) = −( ) = −( ) = −𝑁 ( )
𝜕𝛽 𝜕𝛽 𝑉
𝜕𝛽 𝑉
𝜕𝛽 𝑉
𝑉

Which is clearly the same result as Eq 3.7 A. That is, the mean energy of a
gas composed of 𝑁 indistinguishable molecules is 𝑁 times the mean energy of
a single molecule.

However, in a later Topic when we discuss entropy in detail you will see that it
is important in that case to identify whether the molecules are distinguishable
or not and hence whether Eq 3.10 or 3.11 should be used.

12
3.5. Some important thermodynamic functions
As mentioned above, important thermodynamic quantities can be expressed
as a function of 𝑞 or 𝑄. I’ve listed these below (in terms of 𝑄) and we’ll return
to their application later:

𝑈 − 𝑈(0)
𝑆= + 𝑘ln𝑄 𝐸𝑞. 3.12
𝑇

𝐴 = 𝐴(0) − 𝑘𝑇ln𝑄 𝐸𝑞. 3.13

𝜕ln𝑄
𝑝 = 𝑘𝑇 ( ) 𝐸𝑞. 3.14
𝜕𝑉 𝑇

𝜕ln𝑄 𝜕ln𝑄
𝐻(𝑇) = 𝐻(0) − ( ) + 𝑘𝑇𝑉 ( ) 𝐸𝑞. 3.15
𝜕𝛽 𝑉 𝜕𝑉 𝑇

𝜕ln𝑄
𝐺(𝑇) = 𝐺(0) − 𝑘𝑇ln𝑄 + 𝑘𝑇𝑉 ( ) 𝐸𝑞. 3.16
𝜕𝑉 𝑇

13
Topic 4: Translational partition function

4.1. An expression for the translational partition function 𝒒𝑻

As stated in section 2.3, we use the particle-in-a-box model for a starting point
in considering translational energy levels. We also stated there that the
partition function can be factorised into 3-dimensions as:

𝑇 (𝑋) (𝑌) (𝑍)

𝑞 𝑇 = ∑ e−𝛽𝜀𝑖 = ∑ e−𝛽𝐸𝑛 ∑ e−𝛽𝐸𝑛 ∑ e−𝛽𝐸𝑛 = 𝑞𝑋𝑇 𝑞𝑌𝑇 𝑞𝑍𝑇
𝑖 𝑛(𝑋) 𝑛(𝑌) 𝑛(𝑍)

So, we will initially start by considering a molecule of mass 𝑚 free to move in a

1-dimensional container of length 𝑋. From quantum mechanics we know the
energy levels for such a system, obtained by solution of the Schrodinger
equation are:
𝑛2 ℎ2
𝐸𝑛 =
8𝑚𝑋 2
But remember that we assume that the lowest energy level 𝜀0 is defined as
zero energy, so we need to consider our subsequent energy levels relative to
the 𝑛 = 1 level.
𝑛2 ℎ2 ℎ2 ℎ2
𝜀𝑛 = 𝐸𝑛 − 𝐸𝑛=1 = 2
− 2
= (𝑛2 − 1)𝜀 where 𝜀 =
8𝑚𝑋 8𝑚𝑋 8𝑚𝑋 2

The sum to evaluate to find 𝑞𝑋𝑇 is therefore:

∞

𝑞𝑋𝑇 = ∑ exp (−𝛽(𝑛2 − 1)𝜀)

𝑛=1

We now make two very reasonable assumptions: 1) that translational energy

levels are so close together under typical conditions that we can replace the
sum with an integral (the ‘continuum approximation’) and 2) that extending the
lower limit from 1 to 0 introduces negligible error (and allows the solution of a
standard integral):

∞ ∞

𝑞𝑋𝑇 = ∫ exp (−𝛽(𝑛2 − 1)𝜀 d𝑛 ≈ ∫ exp (−𝛽𝑛2 𝜀) d𝑛

1 0

Which (as shown in worked problems 4) gives the solution:

14
 2𝜋𝑚 1/2
𝑞𝑋𝑇 = ( 2 ) 𝑋 𝐸𝑞. 4.1
ℎ 𝛽

And the translation partition function in 3-dimensions is given by:

3
2𝜋𝑚 2
𝑞 𝑇 = 𝑞𝑋𝑇 𝑞𝑌𝑇 𝑞𝑍𝑇 = ( 2 ) 𝑋𝑌𝑍
ℎ 𝛽

2𝜋𝑚 3/2
𝑇
𝑞 =( 2 ) 𝑉 𝐸𝑞. 4.2
ℎ 𝛽

4.2. Interpretation of 𝒒𝑻

The expression for 𝑞 𝑇 in Eq. 4.2 clearly shows that the partition function
increases with both volume of the container and mass of the molecule. This is
because an increase in each of these factors results in separation between
translational energy levels becoming smaller and hence more of them being
thermally accessible.
Clearly also, as 𝛽 = 1/𝑘𝑇, an increase in temperature results in increased
value of partition function because more energy states become accessible.
Looking back to Topic 2 (section 2.3) where we considered contributions to the
molecular partition function, it was stated that for H35Cl at 298 K, 𝑞 𝑇 ~ 1032
when considering a 1 m3 volume container. You can now see where this
number came from: it was calculated using Eq. 4.2 above (as shown in
Worked Problems 4). Such large values of 𝑞 𝑇 are very typical for even light
molecules in small volumes at room temperature.
Eq 4.2 can be rewritten as:

𝑉 ℎ
𝑞𝑇 = where: Λ = 𝐸𝑞. 4.3
Λ3 (2𝜋𝑚𝑘𝑇)1/2

The quantity Λ has dimensions of length and is called the thermal

wavelength of the molecule. Remembering the approximation that lead to this
expression (the continuum approximation that many states are occupied) we
can understand that Eq 4.3 is only valid if 𝑞 𝑇 is very large. For this to be so,
𝑉 ≫ Λ i.e. the thermal wavelength of the molecule must be far smaller than the
linear dimensions of the container. For H35Cl at 298 K, Λ = 17 pm which is far
smaller than any conventional container (Worked Problems 4).

15
4.3. Mean molecular and internal energy of monatomic perfect gas

We are now going to use our expression for 𝑞 𝑇 to determine thermodynamic

quantities. For simplicity we will consider a monatomic perfect gas, as this
allows us to:
• Consider only translational motion (no rotational or vibrational states).
• Assume the atoms are independent and non-interacting.
Mean molecular energy 〈𝜀〉 is therefore given from Eq 3.4 by:
1 𝜕𝑞 𝑇
〈𝜀〉 = 〈𝜀 𝑇 〉 = 𝜀𝑔𝑠 − 𝑇 ( )
𝑞 𝜕𝛽 𝑉

Which by assuming the 𝜀𝑔𝑠 term can be neglected, and substitution of Eq. 4.2
gives:
3
〈𝜀〉 = 𝑘𝑇
2
The proof of this result is covered in the tutorial questions for this topic.
Further to this Eq. 3.9 and Eq. 3.11 can be used to determine that:
3
𝑈 − 𝑈(0) = 𝑅𝑇
2
4.4. Pressure and Gibb’s energy of a monatomic perfect gas

Equation 3.4 gives an expression for pressure:

𝜕ln𝑄
𝑝 = 𝑘𝑇 ( )
𝜕𝑉 𝑇
Which can be rewritten, using arguments discussed previously:
𝑁𝑘𝑇 𝜕𝑞
𝑝= ( )
𝑞 𝜕𝑉 𝑇
As 𝑞 = 𝑞 𝑇 = 𝑉 ⁄Λ3 (where Λ is independent of 𝑉)
𝑁𝑘𝑇 𝜕(𝑉 ⁄Λ3 ) 𝑁𝑘𝑇 𝑛𝑅𝑇
𝑝= 3
( ) = =
(𝑉 ⁄Λ ) 𝜕𝑉 𝑇
𝑉 𝑉

Which shows very beautifully how statistical mechanics can be related to, and
be consistent with, laws of state such as the perfect gas law.

16
Eq. 3.16 gave an expression for 𝐺:
𝜕ln𝑄
𝐺(𝑇) = 𝐺(0) − 𝑘𝑇ln𝑄 + 𝑘𝑇𝑉 ( )
𝜕𝑉 𝑇
The final term is clearly 𝑝𝑉 (compare to Eq. 3.4) so we can substitute in 𝑛𝑅𝑇
from the result above:
𝐺(𝑇) = 𝐺(0) − 𝑘𝑇ln𝑄 + 𝑛𝑅𝑇
Here, we are considering a gas, where molecules (atoms in this case) are
indistinguishable. We therefore must substitute in Eq 3.11 and if we use
Stirling’s approximation that ln𝑁! = 𝑁ln𝑁 − 𝑁 then we obtain:
𝑞
𝐺(𝑇) = 𝐺(0) − 𝑛𝑅𝑇ln
𝑁
This proof is covered in more detail in Worked Problems 4.
This result allows us to understand Gibb’s energy on a statistical level.
Because 𝑞 represents the number of thermally accessible states and 𝑁 is the
number of molecules, the difference 𝐺(𝑇) − 𝐺(0) is proportional to the log (ln)
of the average number of thermally accessible states available to the
molecules. As 𝑞⁄𝑁 increases, the log (ln) term becomes larger and hence
𝐺(𝑇) − 𝐺(0) becomes more negative. Thus, we can now see that the
thermodynamic tendency to lower Gibb’s energy is driven by the tendency to
maximise the number of thermally accessible states.
Mid-point summary
Hopefully now the relationship between population of molecular energy states
and bulk thermodynamic properties is clear to you as well as the importance of
the partition function in linking the microscopic and macroscopic worlds.
In the lecture material so far, I have explicitly discussed the translational
partition function. This is because at reasonable temperatures translational
motion contributes significantly more to the mean molecular energy of a gas
than vibration and rotation.
The vibrational partition function has also been covered indirectly in the
worked problems and the tutorial questions encourage you to link the relevant
worked problems to vibrational energy levels more explicitly.
In part 2 of this course we will cover the rotational partition function in more
detail, along with other examples, including the electronic partition function.
We will also spend some time considering entropy and the partitioning of
molecules among states from a statistical viewpoint.

17