Lecture 21. Boltzmann Statistics (Ch.

6)
We have followed the following logic:
1. Statistical treatment of isolated systems: multiplicity ⇒ entropy ⇒ the 2nd Law.
2. Thermodynamic treatment of systems in contact with the heat reservoir ⇒ the
minimum free energy principle. However, the link between G and the process of
counting of accessible microstates was not straightforward.
Now we want to learn how to “statistically” treat a system in contact with a heat bath.
The fundamental assumption states that a closed (isolated) system visits every one of
its microstates with equal frequency: all allowed states of the system are equally
probable. This statement applies to the combined system (the system of interest + the
reservoir). We wish to translate this statement into a statement that applies to
the system of interest only. Thus, the question: how often does the system visit
each of its microstates being in the thermal equilibrium with the reservoir? The
only information we have about the reservoir is that it is at the temperature T.

Reservoir R
U0 - ε
Combined system
U0 = const System S
ε

a combined (isolated) system = a heat reservoir and a system in thermal contact

 The Fundamental Assumption for an Isolated System
Isolated ⇒ the energy is conserved. The ensemble of all
σi equi-energetic states ⇒ a mirocanonical ensemble
The ergodic hypothesis: an isolated system in thermal
σ2 equilibrium, evolving in time, will pass through all the accessible
σ1 microstates states at the same recurrence rate, i.e. all accessible
microstates are equally probable.
The average over long times will equal the average over the ensemble of all equi-
energetic microstates: if we take a snapshot of a system with N microstates, we will
find the system in any of these microstates with the same probability.

Probability of a particular microstate of a microcanonical ensemble

= 1 / (# of all accessible microstates)
The probability of a certain macrostate is determined by how many microstates
correspond to this macrostate – the multiplicity of a given macrostate macrostate Ω

Probability of a particular macrostate

= (Ω of a particular macrostate) / (# of all accessible microstates)

Note that the assumption that a system is isolated is important. If a system is coupled
to a heat reservoir and is able to exchange energy, in order to replace the system’s
trajectory by an ensemble, we must determine the relative occurrence of states with
different energies.
 Systems in Contact Reservoir R System S
U0 - ε ε
with the Reservoir
The system – any small macroscopic or
R microscopic object. If the interactions
between the system and the reservoir are
S weak, we can assume that the spectrum of
ε2 energy levels of a weakly-interacting system is
ε1 the same as that for an isolated system.
Example: a two-level system in thermal
contact with a heat bath.

We ask the following question: under conditions of equilibrium between the system
and reservoir, what is the probability P(εk) of finding the system S in a particular quantum
state k of energy εk? We assume weak interaction between R and S so that their energies
are additive. The energy conservation in the isolated system “system+reservoir”:
U0 = UR+ US = const
According to the fundamental assumption of thermodynamics, all the states of the
combined (isolated) system “R+S” are equally probable. By specifying the microstate of
the system k, we have reduced ΩS to 1 and SS to 0. Thus, the probability of occurrence of a
situation where the system is in state k is proportional to the number of states accessible to
the reservoir R . The total multiplicity:

Ω(ε k , U 0 − ε k ) = Ω S (ε k )× Ω R (U 0 − ε k ) = 1× Ω R (U 0 − ε k ) = Ω R (U 0 − ε k )
 Systems in Contact with the Reservoir (cont.)
The ratio of the probability that the system is in quantum state 1 at energy ε1 to the
probability that the system is in quantum state 2 at energy ε2 is:

PS (ε 1 ) Ω R (U 0 − ε 1 ) exp[S R (U 0 − ε 1 ) / k B ] ⎡ S (U − ε ) − S R (U 0 − ε 2 ) ⎤ ⎛ ΔS R ⎞
= = = exp ⎢ R 0 1 ⎥ = exp⎜⎜ ⎟⎟
PS (ε 2 ) Ω R (U 0 − ε 2 ) exp[S R (U 0 − ε 2 ) / k B ] ⎣ k B ⎦ ⎝ B ⎠
k

SR
S(U0) Let’s now use the fact that S is much smaller than R (US=εk << UR).
S(U0- ε1) Also, we’ll consider the case of fixed volume and number of
particles (the latter limitation will be removed later, when we’ll
S(U0- ε2)
allow the system to exchange particles with the heat bath
U0- ε 1
U0- ε 2

UR 1
(dU R + PdVR − μdN R ) (Pr. 6.9 addresses the case
U0

dS R =
TR when the 2nd is not negligible)
0
⎡ ∂S ⎤ ⎡ dS ⎤
S R (U ) ≈ S R (U 0 ) + ⎢ R (U 0 )⎥ (U − U 0 ) S R (U 0 − ε i ) ≈ S R (U 0 ) − ε i ⎢ R (U 0 )⎥
⎣ ∂U R ⎦V , N ⎣ dU R ⎦V , N

⎛ ∂S R ⎞ ⎛ ∂S R ⎞ ⎡⎛ ∂S ⎞ 1⎤
ΔS R = ⎜⎜ (U 0 )⎟⎟ ε 2 − ⎜⎜ (U 0 )⎟⎟ ε1 = ⎢⎜⎜ R
(U 0 )⎟⎟ = ⎥ = − ε1 − ε 2
⎝ ∂U R ⎠V , N ⎝ ∂U R ⎠V , N ⎢⎣⎝ ∂U R ⎠V , N TR ⎥⎦ TR
 Boltzmann Factor
PS (ε 1 ) ⎛ ΔS ⎞ ⎛ ε −ε ⎞ ⎛ ε ⎞
= exp⎜⎜ R ⎟⎟ = exp⎜⎜ − 1 2 ⎟⎟ exp⎜⎜ − 1 ⎟⎟
PS (ε 2 ) ⎝ kB ⎠ ⎝ k BT ⎠ PS (ε 1 ) ⎝ k BT ⎠
=
T is the characteristic of the heat reservoir PS (ε 2 ) ⎛ ε ⎞
exp⎜⎜ − 2 ⎟⎟
⎝ k BT ⎠
exp(- εk/kBT) is called the Boltzmann factor

This result shows that we do not have to know anything about

the reservoir except that it maintains a constant temperature T !

The corresponding probability distribution is known as the

reservoir
canonical distribution. An ensemble of identical systems
all of which are in contact with the same heat reservoir and T
distributed over states in accordance with the Boltzmann
distribution is called a canonical ensemble.
The fundamental assumption for an isolated system has been transformed into the
following statement for the system of interest which is in thermal equilibrium with the
thermal reservoir: the system visits each microstate with a frequency proportional
to the Boltzmann factor.
Apparently, this is what the system actually does, but from the macroscopic point of view
of thermodynamics, it appears as if the system is trying to minimize its free energy. Or
conversely, this is what the system has to do in order to minimize its free energy.
 One of the most useful equations in TD+SP...
Firstly, notice that only the energy difference Δε =
P(ε 1 ) ⎛ ΔS ⎞ ⎛ ε −ε ⎞ εi - εj comes into the result so that provided that
= exp⎜⎜ R ⎟⎟ = exp⎜⎜ − 1 2 ⎟⎟
P (ε 2 ) ⎝ kB ⎠ ⎝ k BT ⎠ both energies are measured from the same origin
it does not matter what that origin is.

P (ε i )
Secondly, what matters in determining the ratio of the occupation
numbers is the ratio of the energy difference Δε to kBT. Suppose = e9 ≈ 8000
that εi = kBT and εj = 10kBT . Then (εi - εj ) / kBT = -9, and P (ε j )

The lowest energy level ε0 available to a system (e.g., a

molecule) is referred to as the “ground state”. If we measure all n(ε ) ⎛ ε ⎞
= exp⎜⎜ − ⎟⎟
energies relative to ε0 and n0 is the number of molecules in this n0 ⎝ B ⎠
k T
state, than the number molecules with energy ε > ε0

Problem 6.13. At very high temperature (as in the very early universe), the proton
and the neutron can be thought of as two different states of the same particle, called
the “nucleon”. Since the neutron’s mass is higher than the proton’s by Δm = 2.3·10-30
kg, its energy is higher by Δmc2. What was the ratio of the number of protons to the
number of neutrons at T=1011 K?
⎛ mn c 2 ⎞
exp⎜⎜ − ⎟⎟
P(n )
= ⎝ k BT ⎠ ⎛ Δmc 2 ⎞
= exp⎜⎜ − ⎟⎟ = exp⎜ −
(
⎛ 2.3 ⋅10 −30 kg × 3 ⋅108 m/s )
2
⎞
⎟ = 0.86
P( p ) ⎛ mpc ⎞ 2
⎝ k BT ⎠ ⎜ 1.38 ⋅10 − 23 J/K × 1011 K
⎝
⎟
⎠
exp⎜⎜ − ⎟
⎟
⎝ k BT ⎠
 More problems
Problem 6.14. Use Boltzmann factors to derive the exponential formula for the
density of an isothermal atmosphere.
The system is a single air molecule, with two states: 1 at the sea level (z = 0), 2 – at
a height z. The energies of these two states differ only by the potential energy mgz
(the temperature T does not vary with z):
⎛ ε ⎞ ⎛ ε + mgz ⎞
exp⎜⎜ − 2 ⎟⎟ exp⎜⎜ − 1 ⎟⎟
P(ε 2 ) ⎝ B ⎠=
k T ⎝ k T ⎠ = exp⎛⎜ − mgz ⎞⎟ ⎛ mgz ⎞
= B
⎜ k T⎟ ρ ( z ) = ρ 0 exp⎜⎜ − ⎟⎟
P(ε 1 ) ⎛ ε1 ⎞ ⎛ ε1 ⎞ ⎝ B ⎠ ⎝ B ⎠
k T
exp⎜⎜ − ⎟⎟ exp⎜⎜ − ⎟⎟
⎝ k BT ⎠ ⎝ k BT ⎠
At home:
A system of particles are placed in a uniform field at T=280K. The particle
concentrations measured at two points along the field line 3 cm apart differ by a
factor of 2. Find the force F acting on the particles.
Answer: F = 0.9·10-19 N
A mixture of two gases with the molecular masses m1 and m2 (m2 >m1) is placed in
a very tall container. The container is in a uniform gravitational field, the acceleration
of free fall, g, is given. Near the bottom of the container, the concentrations of these
two types of molecules are n1 and n2 respectively (n2 >n1) . Find the height from the
k BT ln (n2 / n1 )
bottom where these two concentrations become equal.
h=
Answer: (m2 − m1 )g
 The Partition Function
For the absolute values of probability (rather ⎛ ε ⎞ 1
P (ε i ) = exp ⎜⎜ − i ⎟⎟ = exp (− βε i )
1
than the ratio of probabilities), we need an
explicit formula for the normalizing factor 1/Z: Z ⎝ k BT ⎠ Z
1
β ≡ - we will often use this notation
k BT
The quantity Z, the partition function, can be found from the normalization condition
- the total probability to find the system in all allowed quantum states is 1:

1 = ∑ P(ε i ) = ∑ exp(− βε i ) Z (T , V , N ) = ∑ exp(− βε i )

1 The Zustandsumme
or
i i Z i
in German

The partition function Z is called “function” because it depends on T, the spectrum (thus, V), etc.
∞ ∞
⎛ ε ⎞
Example: a single particle, Z (T ) = ∫ exp⎜⎜ − ⎟⎟dε = k BT ∫ exp(− x )dx = k BT
continuous spectrum. 0 ⎝ k BT ⎠ 0
(kBT1)-1
⎛ ε ⎞ The areas under these curves must be
exp⎜⎜ − ⎟⎟ the same (=1). Thus, with increasing T,
T1
P(ε ) = ⎝ k BT ⎠ 1/Z decreases, and Z increases. At T =
k BT 0, the system is in its ground state (ε=0)
<

(kBT2)-1 T2
with the probability =1.

0 ε
 Average Values in a Canonical Ensemble
We have developed the tools that permit us to calculate the average value of different
physical quantities for a canonical ensemble of identical systems. In general, if the
systems in an ensemble are distributed over their accessible states in accordance with
the distribution P(εi), the average value of some quantity x (εi) can be found as:

x = x(ε i ) = ∑ xi (ε i )P(ε i ) ⎛ εi ⎞
x(ε i ) = ( )
1
In particular, for a
∑ i i ⎜⎜ − k T ⎟⎟
Z (T , V , N ) i
x ε exp
i canonical ensemble: ⎝ B ⎠
Let’s apply this result to the average (mean) energy of the systems in a canonical
ensemble (the average of the energies of the visited microstates according to the
frequency of visits):
U = ∑ ε i P(ε i ) = ∑ ε i exp(− βε i ) = exp(− βε i )
1 1
i i Z Z
∑ε
i
i

The average values are additive. The average total

energy Utot of N identical systems is: U tot = N U

⎛ ε ⎞ ⎡ 1 ⎤ ∂ ∂
Another useful representation for
∑ε i exp⎜⎜ − i ⎟⎟ = ⎢ ≡ β⎥ = −
∂β
∑ exp(− ε β ) = − ∂β Z
i
the average energy (Pr. 6.16): i ⎝ k BT ⎠ ⎣ k BT ⎦ i

1 ∂Z ∂ ∂
U =− =− ln Z = k BT 2 (ln Z ) thus, if we know Z=Z(T,V,N),
Z ∂β ∂β ∂T we know the average energy!
 Example: energy and heat capacity of a two-level system

⎛ 0 ⎞ ⎛ ε ⎞ ⎛ ε ⎞
the slope ~ T Ei
The partition Z = exp⎜⎜ − ⎟⎟ + exp⎜⎜ − ⎟⎟ = 1 + exp⎜⎜ − ⎟⎟
function: ⎝ k BT ⎠ ⎝ k BT ⎠ ⎝ k BT ⎠
ε2= ε
The average energy:

ε1= 0 1 ⎡ ⎛ 0 ⎞ ⎛ ε ⎞⎤ ε
U = ⎜
⎢0 × exp⎜ − ⎟
⎟ + ε × exp⎜⎜ − ⎟⎟⎥ =
Z (T ) ⎣ ⎝ k BT ⎠ ⎝ k BT ⎠ ⎦ ⎛ ε ⎞
- lnni 1 + exp⎜⎜ ⎟⎟
⎝ B ⎠
k T
U 1 ∂Z
U =−
ε /2 (check that the same result follows from
Z ∂β
)

CV

The heat capacity at constant volume:

0
T

⎛ ∂U ⎞ ∂ ⎡ ε ⎤ k B (ε / k BT )2 exp(ε / k BT )
CV = ⎜ ⎟ = ⎢ ⎥=
⎝ ∂T ⎠V ∂T ⎣1 + exp (ε / k B T ) ⎦ [1 + exp(ε / k BT )]2
 Partition Function and Helmholtz Free Energy
∂F ⎛ F 1 ∂F ⎞ 2 ∂ ⎛ −F ⎞
Now we can relate F to Z: U = F + TS = F − T = T 2⎜ 2 − ⎟ = k BT ⎜⎜ ⎟⎟
∂T ⎝ T T ∂T ⎠ ∂T ⎝ B ⎠
k T

⎛ F ⎞
Comparing this with the expression for
Z = exp⎜⎜ − ⎟⎟ = exp(− β F )
the average energy:
⎝ k BT ⎠
This equation provides the connection between the microscopic world which we
specified with microstates and the macroscopic world which we described with F.

If we know Z=Z(T,V,N), we know everything we want to know about the thermal

behavior of a system. We can compute all the thermodynamic properties:

⎛∂F ⎞ ⎛ ∂ ln Z ⎞
S = −⎜⎜ ⎟⎟ = k B ln Z + k BT ⎜⎜ ⎟⎟
⎝ ∂ T ⎠V , N ⎝ ∂ T ⎠V , N

⎛∂F ⎞ ⎛ ∂ ln Z ⎞
P = −⎜⎜ ⎟⎟ = k BT ⎜⎜ ⎟⎟
⎝ ∂ V ⎠T , N ⎝ ∂ V ⎠T , N

⎛∂F ⎞ ⎛ ∂ ln Z ⎞
μ = −⎜
⎜ ⎟
⎟ = k BT ⎜⎜ ⎟⎟
⎝ ∂ N ⎠T ,V ⎝ ∂ N ⎠T ,V
 Microcanonical ⇔ Canonical
Our description of the microcanonical and canonical ensembles was based on counting
the number of accessible microstates. Let’s compare these two cases:
microcanonical ensemble canonical ensemble
For an isolated system, the multiplicity For a system in thermal contact with
Ω provides the number of accessible reservoir, the partition function Z provides
microstates. The constraint in the # of accessible microstates. The
calculating the states: U, V, N – const constraint: T, V, N – const
For a fixed U, the mean temperature T For a fixed T, the mean energy U is
is specified, but T can fluctuate. specified, but U can fluctuate.
E
1 - the probability of finding 1 − k B T - the probability of finding
n

Pn = Pn = e
Ω a system in one of the
Z a system in one of these
accessible states states
S (U , V , N ) = k B ln Ω F (T , V , N ) = −k B T ln Z
- in equilibrium, S reaches a maximum - in equilibrium, F reaches a minimum
For the canonical ensemble, the role of Z is similar to that of the multiplicity Ω for the
microcanonical ensemble. This equation gives the fundamental relation between
statistical mechanics and thermodynamics for given values of T, V, and N, just as
S = kB lnΩ gives the fundamental relation between statistical mechanics and
thermodynamics for given values of U, V, and N.

1 ⎛ ∂S ⎞ P ⎛∂S ⎞ μ ⎛ ∂S ⎞ ⎛∂F ⎞ ⎛∂F ⎞ ⎛∂F ⎞

=⎜ ⎟ =⎜ ⎟ = −⎜⎜ ⎟⎟ S = −⎜⎜ ⎟⎟ P = −⎜⎜ ⎟⎟ μ = −⎜⎜ ⎟⎟
T ⎜⎝ ∂ U ⎟⎠V , N T ⎜⎝ ∂ V ⎟⎠U , N T ⎝ ∂ N ⎠U ,V ⎝ ∂ T ⎠V , N ⎝ ∂ V ⎠T , N ⎝ ∂ N ⎠T ,V
 Boltzmann Statistics: classical (low-density) limit
We have developed the formalism for calculating the thermodynamic properties of the
systems whose particles can occupy particular quantum states regardless of the other
particles (classical systems). In other words, we ignored all kind of interactions between
the particles. However, the occupation numbers are not free from the rule of quantum
mechanics. In quantum mechanics, even if the particles do not interact through forces,
they still might participate in the so-called exchange interaction which is dependent on
the spin of interacting particles (this is related to the principle of indistinguishability of
elementary particles, we’ll consider bosons and fermions in Lecture 23). This type of
interactions becomes important if the particles are in the same quantum state (the same
set of quantum numbers), and their wave functions overlap in space:

strong exchange interaction weak exchange interaction

the de Broglie wavelength λ ↔ (n )1/ 3

the average distance btw particles
h (n ) Boltzmann
1/ 3
h h
λ= ~ (for N2 molecule, λ ~ 10 m at RT)
-11
<< 1 statistics
p k BT m k BT m applies
Violations of the Boltzmann statistics are observed if the the density of particles is very
large (neutron stars), or particles are very light (electrons in metals, photons), or they
are at very low temperatures (liquid helium),
However, in the limit of small density of particles, the distinctions between
Boltzmann, Fermi, and Bose-Einstein statistics vanish.
 Degenerate Energy Levels
If several quantum states of the system (different sets of quantum numbers)
correspond to the same energy level, this level is called degenerate. The probability to
find the system in one of these degenerate states is the same for all the degenerate
states. Thus, the total probability to find the system in a state with energy εi is
⎛ ε ⎞
P (ε i ) = d i exp⎜⎜ − i ⎟⎟ where di is the degree of degeneracy.
⎝ k BT ⎠
⎛ ε ⎞
Taking the degeneracy of energy levels into Z = ∑ d i exp⎜⎜ − i ⎟⎟
account, the partition function should be modified: i ⎝ k BT ⎠
13.6 eV
Example: The energy levels of an electron in the hydrogen atom: ε (ni ) = − 2
ni
ε
r where ni = 1,2,... is the principle quantum number (these levels
εi are obtained by solving the Schrödinger equation for the
di =2ni2
ε2 d2 =8 Coulomb potential). In addition to ni, the states of the electron
in the H atom are characterized with three other quantum
numbers: the orbital quantum number li max = 0,1, ..., ni – 1, the
ε1 d1 =2 projection of the orbital momentum mli = - li, - li+1,...0, li-1,li,
and the projection of spin si = ±1/2. In the absence of the
(for a continuous external magnetic field, all electron states with the same ni are
spectrum, we need degenerate (the property of Coulomb potential). The degree of
degeneracy in this case:
1 + 2(ni − 1) + 1
another approach) l = n −1

∑ [2l + 1] = ni
max i

di = 2 i = 2ni2
l min = 0 2
 Problem (final 2005, partition function)
Consider a system of distinguishable particles with five microstates with energies 0, ε, ε,
ε, and 2ε ( ε = 1 eV ) in equilibrium with a reservoir at temperature T =0.5 eV.
1. Find the partition function of the system.
2. Find the average energy of a particle.
3. What is the average energy of 10 such particles?

⎛ ε ⎞ ⎛ 2ε ⎞
Z1 = 1 + 3 exp⎜⎜ − ⎟⎟ + exp⎜⎜ − ⎟⎟ = 1 + 0.406 + 0.018 = 1.424
⎝ B ⎠
k T ⎝ B ⎠
k T
the average energy of a
single particle: ⎛ ε ⎞ ⎛ 2ε ⎞
ε × 3 exp⎜⎜ − ⎟⎟ + 2ε × exp⎜⎜ − ⎟⎟
ε = ∑ ε i P(ε i ) = ⎝ k BT ⎠ ⎝ k BT ⎠ = 1 eV × 0.406 + 0.036 = 0.310 eV
⎛ ε ⎞ ⎛ 2ε ⎞ 1.424
1 + 3 exp⎜⎜ − ⎟⎟ + exp⎜⎜ − ⎟⎟
i

⎝ k BT ⎠ ⎝ k BT ⎠

3 × exp(− βε )× (− ε ) + exp(− 2 βε )× (− 2ε )
the same
1 ∂Z
result you’d ε =− =−
get from this: Z ∂β 1 + 3 exp(− βε ) + exp(− 2 βε )

the average energy of N = 10 such particles: U10 = N U = 10 × 0.310 eV = 3.1 eV

 Problem 1 (partition function, average energy)
Consider a system of N particles with only 3 possible energy levels separated by ε (let the
ground state energy be 0). The system occupies a fixed volume V and is in
thermal equilibrium with a reservoir at temperature T. Ignore interactions between particles
and assume that Boltzmann statistics applies.
(a) (2) What is the partition function for a single particle in the system?
(b) (5) What is the average energy per particle?
(c) (5) What is probability that the 2ε level is occupied in the high temperature limit, kBT >>
ε? Explain your answer on physical grounds.
(d) (5) What is the average energy per particle in the high temperature limit, kBT >> ε?
(e) (3) At what temperature is the ground state 1.1 times as likely to be occupied as the 2ε
level?
(f) (25) Find the heat capacity of the system, CV, analyze the low-T (kBT<<ε) and high-T
(kBT >> ε) limits, and sketch CV as a function of T. Explain your answer on physical grounds.

(a) Z = ∑ d i exp(− βε i ) = 1 + e − βε + e −2 βε
i

(b)
ε =−
1 ∂Z
=−
(− ε )e − βε + (− 2ε )e −2 βε
=ε
e − βε + 2e −2 βε
Z ∂β 1 + e − βε + e − 2 βε 1 + e − βε + e − 2 βε
(c) e −2 βε 1 − 2 βε 1 all 3 levels are populated with
P= ≈ ≈
1 + e − βε + e − 2 βε 1 + 1 − βε + 1 − 2 βε 3 the same probability

(d) e − βε + 2e −2 βε 1+ 2
ε =ε − βε − 2 βε
≈ ε =ε
1+ e + e 1+1+1
 Problem 1 (partition function, average energy)
2ε
exp(− 2 βε ) =
1
(e) 2 βε = ln 1.1 T =
1 .1 k B ln 1.1
dU d ε
d ε dβ
(f) CV = =N =N
dT dT dβ dT
⎛
= Nε ⎜⎜ −
1 ⎞⎧⎪ (− ε )e − βε + (− 2ε )2e − 2 βε
⎟ −
( )[
e − βε + 2e − 2 βε (− ε )e − βε + (− 2ε )e − 2 βε ]⎫⎪⎬
2 ⎟⎨
⎝ B ⎠⎩
k T ⎪ 1 + e − βε
+ e − 2 βε
1 + e − βε
+ e [
− 2 βε 2
] ⎪⎭
[ ][ ] ( )(
⎛ Nε 2 ⎞⎧⎪ e − βε + 4e − 2 βε 1 + e − βε + e − 2 βε − e − βε + 2e − 2 βε e − βε + 2e − 2 βε
= ⎜⎜ ⎟
)⎫⎪
2 ⎟⎨ ⎬
⎝ k BT ⎠⎪⎩ [
1 + e − βε + e − 2 βε
2
] ⎪⎭
Nε 2 ⎧⎪ e − βε + 4e − 2 βε + e − 2 βε + 4e −3βε + e −3 βε + 4e − 4 βε − e − 2 βε − 4e −3 βε − 4e − 4 βε ⎫⎪
= ⎨ ⎬
k BT 2 ⎪⎩ [
1 + e − βε + e − 2 βε
2
] ⎪⎭
Nε 2 e − βε + 4e − 2 βε + e −3 βε
=
[
k BT 2 1 + e − βε + e − 2 βε 2 ] CV

− βε −2 βε −3 βε ε
N ε e + 4e
2
+e Nε 2 −
CV = ≈ k BT

[ ]
Low T (β>>ε): e
k BT 2 1 + e − βε + e − 2 βε 2 k BT 2
Nε 2 e − βε + 4e −2 βε + e −3 βε 2 Nε 2
high T (β<<ε): CV = ≈
k BT 2
[
−
1+ e + eβε − 2 βε 2
]
3 k BT 2 T
H Problem (the average values)
A gas is placed in a very tall container at the temperature T. The
container is in a uniform gravitational field, the acceleration of
free fall, g, is given. Find the average potential energy of the
molecules.
h

⎛ mgh ⎞
# of molecules within dh: dN (h ↔ h + dh ) = n0 exp⎜⎜ − ⎟⎟(area )dh
⎝ k B T ⎠
0 mgH mgH

⎛ mgh ⎞
H k BT k BT

∫0 mgh × n exp ⎜⎜ − ⎟⎟(area )dh (

∫ B
k T ) y × exp (− y ) ×
k BT
dy ∫ y × exp(− y )dy
⎡ mgh ⎤
0
⎝ k T ⎠ = =
mg
U = ⎥= = k BT ×
B 0 0
⎢ y
H
⎛ mgh ⎞ ⎣ B
k T ⎦
mgH mgH

∫0 n0 exp⎜⎜⎝ − k BT ⎟⎟⎠(area)dh k BT

exp(− y )×
k BT

∫ exp(− y )dy
k BT
∫
0
mg
dy
0

A A

∫ y × exp(− y )dy →
d
[ y × exp(− y )] = exp(− y ) − y × exp(− y ) ∫ dy
0
dy 0
A A A
[ y × exp(− y )] A
0 = ∫ exp(− y )dy − ∫ y × exp(− y )dy ∫ y × exp(− y )dy = − A × exp(− A) + 1 − exp(− A)
0 0 0
A

∫ exp(− y )dy = − exp(− A) + 1

0
For a very tall container (mgH/kBT→ ∞): U = k BT
 Partition Function for a Hydrogen Atom (Pr. 6.9)
Any reference energy can be chosen. Let’s choose ε = 0 in the ground state: ε1=0,
ε2=10.2 eV, ε3=12.1 eV, etc. The partition function:
ε ⎛ εi ⎞
Z = ∑ d i exp⎜⎜ −
(
⎡ 13.6eV 1 − 1 / ni2 ⎤
⎟⎟ = ∑ 2ni exp ⎢−
2 )
r ⎥
i ⎝ B ⎠ i
k T ⎣ k BT ⎦
ε3= -1.5 eV
(a) Estimate the partition function for a hydrogen atom
ε2= -3.4 eV
at T = 5800K (= 0.5 eV) by taking into account only
three lowest energy states.

ε1= -13.6 eV - we can forget about the spin degeneracy – it is the same
for all the levels – the only factor that matters is n2

⎛ ε ⎞ ⎛ ε ⎞ ⎛ ε ⎞
Z = exp⎜⎜ − 1 ⎟⎟ + 4 × exp⎜⎜ − 2 ⎟⎟ + 9 × exp⎜⎜ − 3 ⎟⎟ = e 0 + 4 × e − 20.4 + 9 × e − 24.2 = 1 + 5.5 ⋅10 −9 + 2.8 ⋅10 −10 ≈ 1
⎝ k BT ⎠ ⎝ k BT ⎠ ⎝ k BT ⎠

However, if we take into account all discrete levels, the full partition function diverges:
⎡ ⎛ 1 ⎞⎤
⎢ 13 . 6⎜ 1 − 2 ⎟⎥
∞
⎝ ⎠ ⎛ 13.6 ⎞ ∞ 2
Z = ∑ n exp ⎢− ⎟⎟ ∑ n = ∞
n
2
⎥ > exp⎜⎜ −
n =1 ⎢ k BT ⎥ ⎝ k BT ⎠ n =1
⎢⎣ ⎥⎦
 Partition Function for a Hydrogen Atom (cont.)
Intuitively, only the lowest levels should matter at ε >> kBT . To resolve this paradox,
let’s go back to our assumptions: we neglected the term PdV in

dS R =
1
(dU R + PdVR )
T

⎛ E + PV ⎞
If we keep this term, then New Boltzmann factor = exp⎜⎜ − ⎟⎟
⎝ k BT ⎠

For a H atom in its ground state, V~(0.1 nm)3 , and at the atmospheric pressure, PV~
10-6 eV (negligible correction). However, this volume increases as n3 (the Bohr radius
grows as n), and for n=100, PV is already ~1 eV. The PV terms cause the Boltzmann
factors to decrease exponentially, and this rehabilitates our physical intuition: the
correct partition function will be dominated by just a few lowest energy levels.