You are on page 1of 40

Introduction

to
Statistical
Thermodynamics

1. The Boltzmanns distribution


Given three distinguishable particles, A, B and C.1 Each one of these particles is allowed
to have the energy 0, x, 2x, 3x, etc. Suppose that the total energy (internal energy) of the
system consisting of these three particles is constant and equals 3x. The question now is how this
energy can be distributed over the three particles; or, rephrased, what the various possibilities of
distributing the particles over the energy levels are so that the total energy is 3x. As the following
diagram shows, there are 10 possible ways to achieve the required state. These possible
arrangements are also called complexions. The above required state is thus represented by ten
complexions. It is important to notice that the energy of any given particle is, as a result of the
continuous collisions between these particles and the accompanying energy exchange, not
constant with time; the total energy, however, is (isolated system).
A

3x

2x
ABC
B C

A B

A C

1x

n3=0
n2=0
n1=3
n0=0

n3=1
n2=0
n1=0
n0=2

n3=0
n2=1
n1=1
n0=1

The fundamental assumption of statistical thermodynamics is that all possible


complexions are equally probable. In other words, there is no preference of one arrangement over
the other. Any particle is likely to have one energy value as the other.
1

For example, atoms of a solid are distinguishable because the atoms are localized and each atom is characterized by
its position, unlike the atoms of a gas.

The 10 complexions shown above represent three micro-states X, Y and Z. Each microstate is defined by its own population distribution (i.e. the ni profile, where ni is the number of
particles occupying the ith energy level). Consequently, one complexion corresponds to the microstate X, three complexions correspond to Y and six complexions correspond to Z. We conclude
therefore that it is more probable to find the system in the micro-state Z than in the micro-states
X or Y. In general, the higher the number of complexions corresponding to one state is, the more
probable is this state.
Our next task is to derive a general formula to calculate the number of complexions
corresponding to a specific state, i.e. that with n0 particles in the energy level 0, n1 particles in the
energy level 1, n2 particles in the energy level 2, and so on. The above problem is identical with
having a sac containing N distinguishable balls (e.g. of different colors). The balls are to be
distributed in a certain number of boxes so that n balls come in the first box, n come in the
second box, n in the third box, and so on. The number of different possibilities achieving that
state can be shown to be given by
W

N!
N!

n0 !n1! n2 !... ni !

(1.1)

Notice that the number of possibilities has been divided by the number of permutations of
the particles in a certain energy level among themselves because the order of pulling out the balls
is not important. To elucidate this point further consider the state X in the system described
above. The first energy level contains three particles A, B and C. It is not important whether A
then B then C were pulled out of the sac or A then C then B or B then C then A etc.1 The six
possibilities (see footnote 1) reduce thus to a single complexion! Applying the above equation
for the states X, Y and Z gives2
WX

3!
1
0!3!0!0!

WY

3!
3
2!0!0!1!

WZ

3!
6
1!1!1!0!

Equation (1.1) has very important consequences as the number of particles N grows very
large. Suppose the system contains one mole of particles. The number of possible arrangements
(complexions) of having these Nav particles distributed over Nav energy levels (i.e. a single particle
per energy level, as in the Z state) is Nav!. The number of possible arrangements (complexions) of
1
2

These are six possibilities ABC, ACB, BAC, BCA, CAB and CBA.
0!=1.

having all these Nav particles in the same energy level (as in the X state) is 1. The chance of
finding the system in the X state is now practically zero (compared to 10% when the total number
of particles was just 3). In other words, the system is expected to be found entirely in the state
with maximum number of complexions (the Z state in this case). This state is not just the most
probable; it is the stable state that can only be formed. It is the state of equilibrium. Small
fluctuation from this state may occur, however with less probability. Suppose that one of the Nav
drops from its energy level to the one below. As a result another particle must jump from its
energy level to the one above in order to keep the energy constant. The probability of finding the
system in this state is W=N!/4. If two particles drop to a lower energy level and two jump to a
higher one, then W=N!/8.
The question now arises whether it is possible to determine the most probable state (the ni
profile over the energy levels). Mathematics delivers the answer. We are looking for the
maximum value of W which is again the maximum value of lnW.1 The first derivative must thus
be zero. We make also use of the fact that W is a state function having an exact differential.
W f (n1 , n2 , n3 , n4 ,...) max
ln W f (n1 , n2 , n3 , n4 ,...) max
ln W
n1

d ln W

ln W
n2

dn1

ln W
n3

dn2

dn3 ... 0

Two constraints now apply:

N n1 n2 n3 ... ni const.
dN d n1 d n2 d n3 ... dni 0
E 1n1 2 n2 3 n3 ... i ni const.
dE 1d n1 2 d n2 3 d n3 ... i dni 0
According to Lagrange, we introduce the undetermined multipliers and . Multiplying
the two constraints with and , respectively, and adding dlnW leads to equation 1.2.

d ln W dni i dni 0
Equation 1.2 is actually an equation system:

It is easier to manage the problem in the logarithmic form.

(1.2)

ln W

n1

dn1 dn1 1dn1

ln W
n2

ln W
n3

dn2 dn2 2 dn2

dn3 dn3 3 dn3

............

which can be summarized as


ln W

i d 0
ni

(1.3)

In the above equation system, there are i variables with two conditions (constant N and
constant E). That means there are i-2 independent variables. Now let us choose and so that
ln W

n1

ln W

n2

dn1 dn1 1dn1 0

dn2 dn2 2 dn2 0

The equation system 1.3 reduces thus to


ln W

i d ni 0
i2
ni

Since these i-2 variables are independent, then changing one of them doesnt affect the
others. This means that the sum of all terms that involve that specific variable i must,
independently of the others, be zero, i.e.:
ln W
i 0
ni

(1.4)

Well try now to evaluate (dlnW/dni) and will thereby make use of the Stirlings theorem:
1
1

ln N ! N ln N N ln 2
2
2

If N is very large, the above formula reduces to

ln N ! N ln N N
Applying the Stirlings theorem in equation 1.1 yields

N!
n0 !n1! n2 !...

ln W ln N ! ln n0 ! ln n1! ln n2 ! ...

ln W N ln N N n0 ln n0 n0 n1 ln n1 n1 n2 ln n2 n2 ...
ln W N ln N ni ln ni N n0 n1 n2 ...

and
W
1
ni ln ni 1 ln ni
ni
ni

(1.5)

Substituting equation 1.5 in equation 1.4 gives

ln ni i 0
ln ni i
ni e e

Since a is constant, ea is constant and is set equal to A:

ni Ae

(1.6)

Equation 1.6 is the so-called Boltzmanns distribution. This is the distribution of the most
probable state which represents the equilibrium distribution. Fluctuations from this state are
limitingly small.
If the ith energy level were degenerate with the degeneracy gi, equation 1.6 can be shown
to become

ni Ag i e

(1.7)

Equation 1.7 can be applied also even if the energy states are not exactly degenerate but
are very close to each other and fall within the energy range of d. In such case, one speaks of
energy bundles of d-width and gi is the number of energy states within this bundle. This is very
useful when treating continuous energies as will be shown in the case of translational energy.
With help of equation 1.6 or equation 1.7, the fraction of particles with the energy i can
be determined (this is the probability of finding the particle in the ith energy level):

ni
ni
Ae i
Ae i
e i

N ni Ae i A e i e i
i

ni
ge
i i
N gie

or

(1.8)

and the ratio of particles in the ith energy level to those in the jth energy level is given by

ni e

e
,
nj e

ni g i e g i
e
nj g je
gj

The summation

e q ( g e
i

(1.9)

) in the above expressions is called the partition

function and is given the symbol q. This function is very important in statistical thermodynamics.
When known, all thermodynamic properties of the system can be calculated. More to this issue
and to the physical meaning of the partition function is found in section ??.
The constant A in the Boltzmanns distribution (eq.1.6 or 1.7) is determined as follows:
N ni Ae i A e i A q
A

N
q

2. Entropy and disorder


As explained in section 1, the most stable state is that with maximum number of
complexions. On the other hand and according to the laws of thermodynamics, the equilibrium
state is characterized by maximum entropy. It seems therefore plausible to assume that entropy is
associated with the number of complexions. In the language of mathematics,
S f (W )

Let us now consider two subsystems A and B:

Stotal S S A S B

Wtotal W WA WB

f (W ) f (WA ) f (WB ) f WA WB

Differentiating with respect to WB gives:

f WB WA f WA WB
Differentiating with respect to WA gives:
0 W W f W W f W W
f W W W f W W
f W
1

f W
W
ln f W ln W const .
A

k
W
k ln W W0

f W
f

S k ln W S 0

For conveniency, the constant W0 is set to be zero and we are left with the famous
equation for statistical entropy engraved on Boltzmanns tombstone in Vienna:

S k ln W

(2.1)

Above equation relates entropy with disorder. The larger the number of complexions
corresponding to a given state, the larger is the disorder in this state. Disordered states are more
stable because they can be achieved in more possible ways than ordered ones. The constant k in
equation 2.1 is the so-called Boltzmanns constant. Interestingly, Boltzmann himself didnt
determine its value. It was Planck who first made in 1900 an estimation of this value from his
solution of the black body radiation problem.
8

By correlating statistical mechanics with thermodynamics, the value of can be


determined:
ln W
i 0
ni

ln W i ni 0
ln W i ni

d ln W ln W i dni
d ln W dni i dni

Suppose now that the system is supplied slowly with heat. The number of particles will
not thereby change (i.e. dni=0) but their distribution on the various i energy levels will. The
heat supplied to the system (dQ) is then equal to the change in the system energy (idni).
Substituting in the above expression yields thus:
d ln W 0 dQ
d ln W

dQ

The definition of entropy in thermodynamics is given by

dS

dQrev
T

Then, by applying equation 2.1


dQrev Tds Tkd ln W
d ln W
1

dQrev
kT

Comparing the two expressions for dlnW/dQ gives


1
kT

and the Boltzmanns distribution now reads

ni Ae

kT

(2.2)

Well try now to evaluate the Boltzmanns constant k. Consider a particle in threedimensional box with the dimensions a, b and c. The energy of such particle is given by quantum
mechanics to be

pqr

h2 p 2 q 2 r 2

8m a 2 b 2 c 2

p, q and r are thereby positive integers (quantum numbers). Applying the above
expression for energy in the Boltzmanns distribution gives

N ni Ae
N Ae

Ae

i kT

h 2 p 2
8 mkT a 2

h 2 p 2 q 2 r 2

8 mkT a 2 b 2 c 2

h 2 q 2
8 mkT b 2

h2 r2

8 mkT c 2

All constants in the exponent can be reduced to a single constant . Thus,

h 2 p 2
8 mkT a 2

p2

h
a 8mk T

Because the energy levels are very close to each other, the summation can be replaced by
integration. The resulting integration is a standard one.

p2

e p dp
2

2 m kT
2 h

The same procedure is applied over the other two summations over q and r. The total
number of particles is then given by
NA

abc

2 mkT 3 2 A V 2 mkT

3
2
h
h

32

From Boltzmanns distribution


N ni A e

dN
dN
A i e E
d
d 1 / kT
i

2 m
N A V 2
h

dN
3
2 m
A V

2
d
2
h

32

32

5 / 2

3
N kT E
2

From the kinetic gas theory


2
E
3
pV N k T
pV

Now comparing with the ideal gas law (pV=nRT) yields

10

N k T n RT
nR
R
k

kB
N
N av

The Bolzmanns constant is thus the general gas constant per particle.

11

3. The partition function


3.1 The molecular partition function and its interpretation

It is a sum of Boltzmann factors,

kT

, that specify how the particles are partitioned

throughout the accessible states. The numerical value of the partition function represents the
effective number of energy levels thermally accessible for the particle. To elucidate this point,
consider a two-level system1; the lower energy level has the energy zero and the second energy
level has the energy 1. The partition function reads

q e
i

kT

e 0 e

kT

1 e

kT

At very low temperatures (T0), the second term approaches zero (e-) and q equals 1.
This means that the particle only exists in the first energy level. As the temperature is increased,
the second term increases and the value of q increases. A q-value higher than 1 means that the
particle has now also access to the second energy level but with lower probability than for
existing in the first energy level. As T approaches ,

kT

goes to unity and q =2. This means

that the particle can occupy the two levels with equal probability.
1.40

2.0

1.35
1.8

1.30
1.25

1.6

1.20
1.4

1.15
1.10

1.2
1.05
1.00

1.0

0.95
0

10

0.0

0.2

0.4

0.6

0.8

1.0

kT/1

kT/1

Another important example is a system with infinite number of equidistant energy levels
at 0, , 2, 3, etc, where is the energy difference between two adjacent energy levels.

An example of two level system is the electronic spin energy in presence of external magnetic field that leads to
lifting of degeneration.

12

This resembles the vibrational energy levels of harmonic oscillator. The corresponding partition
function q reads

q e
i

kT

1 e

kT

kT

kT

...

q 1 e kT e 2 kT e 3 kT ...
q 1 e a e 2 a e 3 a ... 1 e a e a e a ...
1
1
1
q 1 x x 2 x 3 ...

a
1 x 1 e
1 e kT
1

10

12

10

0
0

kT/
The number of thermally accessible levels depends clearly on the ratio kT/ where kT
represents the thermal energy supplied by the surroundings. The higher the thermal energy (the
higher the temperature), the more are the levels that the molecule can exist in. Note that as the
energy level increases, the probability to exist at that energy level decreases (Boltzmanns
distribution).
The vibrational energy of harmonic oscillator differs slightly from the ladder system
described above since the energy of the zeroth level is not zero. According to the laws of
quantum mechanics the energy of the zeroth vibrational level is the zero point energy which is
equal to 1/2ho, where o is the fundamental frequency of oscillation. The Boltzmanns
distribution requires however that the energy of the zeroth level is zero. In order to apply the
Boltzmanns distribution to the vibrational energy of harmonic oscillator, the vibrational energies

13

must be shifted donwards so that o becomes zero. This is achieved by subtracting the zero point
energy from the actual vibrational energies. Taking in consideration that the energy difference
between any two adjacent energy levels of the harmonic oscillator is ho

q e
i

kT

1 2 h 0

kT

e 1 2 h
1

kT

1 2 h 0

kT

1 2 h 0

kT

...

q 1 e kT e 2 kT e 3 kT ...
q 1 e h

kT

e 2 h

kT

e 3 h

1
1 e h

kT

...

kT

(3.1)

Exercise: Calculate the fraction of 1H35Cl (o=2886 cm-1) molecules present in the zeroth
vibrtional energy level at room temperature and at 1000C. Do the same calculation for the 127I35

Cl molecules (search for the fundamental frequency!). Compare the results and explain the

observed variation.
The partition function of rotation is more difficult to compute. For a linear molecule, the
energy of rotation is given by

J J ( J 1)

h2
J ( J 1) h B
8 2 I

14

where I is the moment of inertia, B the rotational constant and J the rotational quantum number
with values 0, 1, 2, etc. The rotational levels show 2J+1-fold degeneracy. The partition
function thus reads

q g i e

2 J 1 e J ( J 1) hB kT

kT

(3.2)

J 0

Given the value of B, the partition function can be evaluated numerically (i.e. for each
value of J, the term is calculated, the terms are then summed up; you will see that the series
above converges to a certain value).
Exercise: Use Excel or Origin software to evaluate the rotational partition function of 1H35Cl at
25C. B=10.591 cm-1.
When the thermal energy kT is much larger than the energy difference between two
neighboring rotational levels, the sum in equation 3.2 can be approximated by an integral.
Equation 3.2 becomes
q g i e

kT

2 J 1 e J ( J 1) hB kT 2 J 1 e J ( J 1) hB kT dJ
J 0

kT d J ( J 1) hB kT
kT J ( J 1) hB kT
e
e
0 kT 0 1

dJ

h B 0 dJ
hB
hB

kT
hB

(3.3)

Exercise: Evaluate the rotational partition function of 1H35Cl at 25C using equation 3.3.
To derive an expression for the partition function of translation per degree of freedom, we
consider the particle to behave as a particle in a box. Its energy is thus given by

h2 p 2
p

8m a 2
where a is the box length and p is a positive integer. The partition function can be approximated
by an integer because the energy levels are very close to each other (continuum).

q e
with

i kT

e
p

h2 p2
8 m a 2 kT

e
0

h
,
a 8mk T

15

h2 p2
8 m a 2 kT

dp e p dp
0

The quantity

2 m k T

a
2 m kT
h

(3.4)

is given the symbol and called the thermal wavelength. For

three dimensions;
abc
32
2 mkT
q 3 2 mkT V

h
h2

32

V
3

Notice that the partition function of translation depends on the size of the container.
Exercise: Calculate the translational partition function of an H2 molecule confined to 100 cm3 at
25C.
In the following it is shown that the total partition function of a molecule is the product of
individual partition functions. Assuming that the various energy contributions of a molecule are
independent (Born-Oppenheimer Approximation):

total electronic translational rotational vibrational


q e

kT

q e
ielec 1

electronic

kT

electronic

translational rotational vibrational

translational kT

itrans 0

rotational kT

J 0

kT

vibrational kT

v 0

q qelectronic qtranslational qrotational qvibrational

(3.5)

The same procedure can be used to show that the total partition function of any energy
contribution is the product of the individual partition functions per degree of freedom:
qtranslational q xtrans q ytrans q ztrans
qrotational q xrot q yrot q zrot

nonlinear

qrotational q x q y

linear

rot

rot

q vibrational q1vib q 2vib q3vib ...q3vibN 6

nonlinear

q vibrational q1 q 2 q3 ...q3 N 5

linear

vib

vib

vib

vib

(3.6)

(3.7)

3.2 Significance of the partition function


The importance of the molecular partition function lies in the fact that it contains all
information needed to calculate all macroscopic thermodynamic properties of systems of

16

independent particles1,2. Since the molecular partition function is related to a single molecule, socalled system partition functions Q are defined for systems containing N particles (as the case is
in real systems). It can be shown that for a system containing N distinguishable independent
particles Q=qN. For N indistinguishable independent particles, Q=qN/N! because permutations of
the particles among themselves have to be counted.
a) Internal Energy

E ni i A i e kT
N A e
A i e kT e kT N
E
i e kT
kT
q
e
i

kT

d ln e
dT

i
1 i e kT

kT
kT 2 kT 2 e kT
d ln e kT
1 A i e kT
1 E
2
2
kT
dT
k T A e
kT N
d ln q
1 E
2
dT
kT N
N
d ln Q
2 d ln q
2 d N ln q
2 d ln q
E N kT
k T
kT
kT 2
dT
dT
dT
dT
i

kT

kT

Taking in consideration that the internal energy at zeroth level is not necessarily
zero:
d ln q
dT
d ln q
U (T ) U (0) N k T 2
dT
E U (T ) U (0) N kT 2

b) Entropy

S k ln W k ln

N!
k ln N ! ln ni !
ni !

S k N ln N N ni ln ni ni

For systems of interacting molecules, the so-called canonical partition functions are used but this is out of the scope
of this introduction.
2
In this sense, it resembles the wave function in quantum mechanics that contains all information about the quantum
system. Partition functions are thus some sort of thermal wave functions

17

with

ln ni i

n N

S k N ln N ni i k N ln N ni ni i
N
, ni i E , ni N
q
S k N ln N ln N ln q N E k N ln q k E
1
with

kT
E
S k N ln q
T
d ln q
d ln q

S k N ln q N k T
k N ln q T

dT
dT

d ln Q

S k ln Q T

dT

with

A e

c) Helmholtz free energy


E
T
T S k N T ln q E
k N T ln q E T S A
A k N T ln q
S k N ln q

A A(0) k N T ln q kT ln Q

d) Pressure
dA SdT pdV
A

ln q

ln Q

k NT
T

kT
T

e) Free Gibbs energy


G H TS E pV TS A pV
ln q

G k N T ln q k N T V

ln q

G G (0) k N T ln q k N T V

ln Q

G G (0) k T ln Q k T V

The free Gibbs energy of indistinguishable ideal gas particles requires special attention.

18

G A pV
G G (0) k T ln Q pV k T ln Q n R T
qN
n R T k T ln q N kT ln N ! n RT
N!
G G (0) k N T ln q k T N ln N N n R T
G G (0) k T ln

G G (0) k N T ln q k N T ln N k N T n R T
with

N
N av

R k N av

and

G G (0) k N T ln q k N T ln N n RT n R T
q
G G (0) k N T ln
N

The last equation gives a new interpretation for the free Gibbs energy. It is proportional to
the logarithm of the number of thermally accessible states per molecule. Introducing the molar
partition function qm when N=Nav yields

qm
N av

Gm Gm (0) RT ln

(3.8)

Exercise: Calculate the entropy of a collection of N independent I2 molecules (o=214.6 cm-1) at


25C assuming harmonic oscillator behavior.

1
1 e h kT
ln q ln(1 e h
q

d ln q
d ln(1 e h

dT
dT
d ln q h 0
1
2 h kT
dT
kT e
1

kT

1 e h

kT

kT

h 0 kT

h 0
h 0 e h kT
2
kT 2
kT 1 e h kT
0

d ln q

h
S k N ln q T
k N ln 1 e
dT

19

kT

kT e

1
h 0 kT

30

25

-1

S / Jmol K

-1

20

15

10

0
0

10

kT/h 0
For I2 at 25C:
kT
1.38 10 23 J K 1 298 K

0.968
h o 6.6 10 34 J s 214.6cm 1 3.0 1010 cm s 1

From graph, S = 8.4 Jmol-1K-1.


Exercise: Evaluate the molar entropy of N two level systems and plot the resulting expression.
Exercise: Calculate the vibrational contribution to entropy of Br2 at 600 K given that the
wavenumber of the vibration is 321 cm-1. Calculate the vibrational energy of the system.

20

Collision
Theory

21

4.1 Translational Energy Distribution


In this section, the Maxwell-Boltzmann distribution is to be derived. But first of all, some
thoughts about the Planck's constant h have to be considered. The unit of h is J s. In basic SI units
J s Nms

kg m 2
m
s kg m
2
s
s

which is equal to the unit of linear moment p=m.v (kgm/s) multiplied by the unit of position x. h,
thus, has the dimension of moment distance (px).
Let's consider now the so-called phase plane for one dimensional motion1. This is the

Linear moment , p

plane constructed by plotting the linear moment of motion versus the position.

dp

h
dx
position, x

The smallest unit in such a plane is h (the cyan rectangle). Each discrete state (defined by
the values of p and x) is represented by such a rectangle in the phase plane, corresponding to an
energy i=px2/2m.
Next step is to determine the degeneracy gi of gaseous molecules. These are the energy
states that are so close to each others and lie within the energy interval d. Obviously, this is the
number of states (rectangles) belonging to that bundle represented by the rectangle dpdx. Thus,
1

For higher dimensions we speak of phase space which is, as an example, a six dimensional space in the case of
three dimensional motion (x,y,z,px,py,pz).

22

gi

dx dpx
h

For three dimensional motion:


gi

dx dpx dy dp y dz dp z

h
h
h

Applying equations 1.8 and 3.4, taking thereby in consideration that the separation
between the energy levels of translational motion are so small that the summation can be replaced
by integration
dN i x, p x e i kT dx dp x

N
q
h

Ni
g i e i

N g i e i
i

dN i x, p x
dx dp x
e i kT
e mv x 2 kT

m
dx dv x
a
N
h
a 2 mkT
2 m kT
h
E pot 0 (ideal gas)
dp d (mv) mdv
2

a
dN i x, p x
e mv x 2 kT
e mv x 2 kT
m
dv x dx m
dv x
N
a 2 m kT
2 mkT
0
2

The last equation above gives the fraction of particles with kinetic energy i located
between 0-a irrespective of position. It is usually written in the form
1 dN i x, p x
e mv x 2 kT
m
N
dv x
2 m kT
2

(4.1)

`The velocity distribution according to equation (4.1) is shown below for N2 molecules at
298 K and 1500 K. The y-axis represents thereby the fraction of particles with velocity vx. Note
that the curve is symmetrical and extends to infinity in both directions (the direction of the one
dimensional motion is thereby taken in consideration). The average speed is zero because motion
in opposite directions cancels each other. As the temperature is increased, the curve broadens
since higher temperatures means higher kinetic energies which in turn means higher velocities.
The curve center remains at zero because no direction is preferred over the other. Notice the
absolute value of the speed with which the molecules move.

23

For three dimensions,


abc
2 mkT 3 2
h3
v 2 vx2 v y2 vz2
q qx q y qz

dN i v x , v y , v z
N

2
m
e mv x
2 mkT

2 kT

m
mv 2
e y
2 mkT

dv x

dN i v x , v y , v z
m

N
2 kT

2 kT

2
m
e mv z
2 mkT

dv y

2 kT

dv z

3 2

e mv

2 kT

(4.2)

dv x dv y dvz

2
Transfering equation 4.2 into spherical coordinate system ( dv x dv y dv z 4 v dv ) gives

dN i vx , v y , v z

m
4 v 2
N
2 kT

(4.3)

24

3 2

e mv

2 kT

dv

Equation 4.3 represents now the fraction of particles with speed v irrespective of
direction. The volume element over which integration is carried out (4v2dv) is a very thin shell
around the origin with thickness dv and radius v.

25

The average speed of gas particles can be evaluated according to

v dN (v)

v dN (v)
0

dN (v)
0

Substituting equation 4.3 gives


v

3 2

m
2

kT

2
v 4 v

e mv

2 kT

m
d v 4
2

kT

3 2

v e
3

mv 2 2 kT

dv

with

x e
3

x2

dx

1
2 2

and
3 2

m
2

kT

v 4

m
2kT

4k 2T 2 8 k T


2m 2
m

12

(4.4)

The Maxwell-Boltzmann distribution (equation 4.3) can also be written in terms of the
kinetic energy of gas particles:
E

mv 2
2

dE
12
mv 2m E
dv

Equation 4.3 becomes


dN i
2
m

4
N
m 2 kT

3 2

d
2

1 2e
12
3 2
2m
kT

kT

The average energy can be shown to be

dN ( )
0

dN ( )
0

dN (v)
0

26

3
kT
2

kT

(4.5)

4.2 Rate of reaction


The basic idea in collision theory is simple and intuitional. For a chemical reaction to
occur, the reactant molecules must "meet" or "collide". They must encounter at a distance short
enough to enable energy exchange. Let us consider the following bimolecular reaction to
elucidate these points.

As long as the A2 and B2 molecules are kept in separate containers (e.g. beakers), no
chemical change will take place. Mixing the two reactants together is a prerequisite for the
reaction to occur because it gives the reactant molecules the chance to "meet". However mere
colliding is not enough. As can be concluded from the reaction equation above, chemical
reactions are processes in which bonds are redistributed, i.e. present bonds are broken and new
are formed. But breaking bonds requires energy and the collision theory states that the required
energy can be obtained from the energy exchange in the collision. The colliding pair must thus
have energy sufficient to overcome such a barrier. If not, the collision is said to be non-reactive.
According to the above concept, the rate of reaction must be proportional to the rate of
collision. We'll show below how this is done for bimolecular collisions of gas particles A and B.
Imagine a gas phase system containing the particles A and B as shown schematically in the figure
below. The A particles are assumed to be stationary whereas the B particles move horizontally
along the drawn cylinder. The question to be answered now is: with which A particles is this
specific B particle going to collide. A simple geometric consideration reveals that collisions occur
only if the separation d between the centers of A and B particles is not larger than the sum of the
radii of A and B (d rA+rB). The cross-sectional area of the hypothetical cylinder confining all A
particles with which the B particle is going to collide is called the cross-section of collision and
is given by
2
d AB
rA rB

The next question is about the number of collisions that this B particle is going to make in
one second if it moves with an average speed vB . This is the question about the number of A
particles the center of which is confined in the above described hypothetical cylinder with length

27

x, which is the distance moved by particle B in one second. Obviously, this depends on the
concentration of A. We define NA, the number of A molecules per unit volume. The number of A
particles confined in the cylinder with length x is given by
Vcylinder N A x N A vB N A
x v t

t 1s

xv

The expression above gives the number of collisions of one B particle with A particles in
one second. Considering that there are NB particles per unit volume, the following equation for
the number of collisions of A and B, ZAB, in one second and per unit volume is given:
Z AB vB N A N B

28

In the above derivation, the A particles were assumed to be stationary. This, however, is
not the case. This is why in the above expression the average velocity of B relative to A must be
substituted instead of the actual velocity of B.
Z AB vBrelative N A N B

(4.6)

2
Z AB d AB
vBrelative N A N B rA rB vBrelative N A N B
2

vrel

For a 90-collision between A and B particles, the average velocity is given by


vrel v A2 vB2
8k T 8 k T


vrel
mA mB

8k T

8k T
1
1


mA mB

1/ 2

Equation 4.6 becomes


2 8k T
Z AB rA rB

1/ 2

(4.7)

N ANB

For a 90-collisions between like molecules, the average velocity is


vrel

2 vA

Equation 4.6 becomes


Z AA rA rA

8k T
2
mA

1/ 2

kT
N
N A A 2 d A2
2
mA

1/ 2

N A2

(4.8)

However, not all these collisions (given by equations 4.7 or 4.8) are reactive. A certain
energy barrier, b, must be overcome, i.e. only collisions with b are reactive:

29

In the simplest form of the collision theory where only two dimensions are considered
(collision takes place in a plane), the fraction of particles fulfilling the above condition can be
determined from the Maxwell-Boltzmann distribution (equation 4.5 but for two dimensions)1:
dN

0
Ni k T 2 2 e
b

dN
1
Ni k T e
b

kT

kT

1
k T e
kT

kT
b

e b

kT

The number of reactive collisions is thus:


2 8k T
*
Z AB
rA rB

*
AB

8k T

1/ 2

N A N B e b

kT

(4.9)

1/ 2

N ANB e

b kT

Now let's consider the elementary reaction


A + B product
rate k A B k
rate
Z

The general equation is

*
AB

N A NB
N av N av

d A
1 d NA
N N

k A B
dt
N av dt
N av N av

8k T
d NA
N

k N A B
dt
N av

dN
1
s 1
Ni k T s e
b

1/ 2

N A N B e b

where 2s equals the number of dimensions.

30

kT

kT

8k T
k

1/ 2

N av e b

8k T

k rA rB

2

kT

(4.10)

1/ 2

N av e

b kT

The constants in the pre-exponential term in equation 4.10 are summarized as BAB:

k B AB T 1 / 2 e b

kT

(4.11)

Equation 4.11 explains the temperature dependence of the rate constant. As temperature is
increased, the fraction of particles with b increases as dictated by the Maxwell-Boltzmann
distribution.

Equation 4.11 predicts also a temperature dependence of the pre-exponential factor. In the
following a comparison with Arrhenius equation is conducted:
k Aobs e E a / RT
ln k ln Aobs
d ln k Ea
2
dT
RT

k BAB T 1 / 2e Eb / RT
Ea
RT

1
E
ln k ln BAB ln T b
2
RT
d ln k 1
E
1 / 2 R T Eb

b2
dT
2 T RT
RT 2

Ea 1 / 2 R T Eb

If Eb >>Ea, then Eb Ea. Under such conditions, Aobs=T1/2BAB. The latter is often used to
check the agreement of the kinetic values predicted by collision theory with those experimentally
observed. The table below summarizes such a comparison for a selected set of gas phase

31

reactions. As can be seen, the agreement is good only in simple cases (reactions #1 and #5). One
suggestion to explain the deviation of experimental values from those calculated by the collision
theory was based on the idea that in addition to energy requirements there are some steric
requirements for a collision to be reactive. The steric requirements are clearly shown in the
figure below.

log

Aobs
T 1/ 2

log BAB

H2+I2 2HI

6.78

7.38

NO + O3 NO2 + O2

4.80

6.90

D + H2 DH + H

5.99

7.32

CH3 + H2 CH4 + H

4.25

7.27

CH3 + CH3 C2H6

6.32

6.62

C2H4 + butadiene cyclohexene

2.80

7.20

32

To take these steric requirements into account, the so-called steric factor p must be
inserted in equation 4.11. It takes values between 0-1. p=1 means that no such requirements exist
and any collision with b is reactive. p=0 means that all collisions regardless of their energies
are non-reactive; the steric requirements are so high that they are never fulfilled.

k p BAB T 1 / 2 e b

kT

(4.12)

Experimental rate constants compared to the ones predicted by collision


theory for gas phase reactions
Reaction

ZAB

Steric factor

2ClNO 2Cl + 2NO

9.4 109

5.9 1010

0.16

2ClO Cl2 + O2

6.3 107

2.5 1010

2.3 10-3

H2 + C2H4 C2H6

1.24 106

7.3 1011

1.7 10-6

Br2 + K KBr + Br

1012

2.1 1011

4.3

Special attention should be paid to the last reaction in the above table. The steric factor is
much larger than 1. The above reaction is an example of the so-called harpoon reactions. When
the K atom is close enough to the bromine molecule, an electron (the harpoon) flips from the K
atom to the bromine molecules forming ions. The ions attract each other producing KBr. The
collisional cross section is thus largely increased in comparison to the geometric ones.This means
that the collisional cross section is much larg
However, with the introduction of the steric factor, the collision theory has lost its
prediction power since the steric factor can not be computed. However, the simple pictorial
model with intuitive concepts keeps reserving an important place for the collision theory in
chemical kinetics.
Reaction probability: Energy exchange between particles depends on the way collisions
occur. In the derivation of equation 4.10, it was assumed that the collisions are head-on or central
collisions. This is not always true. Consequently, collisions are reactive only if the energy
component in the direction of collision (and not the total energy) is larger than the barrier. The

33

magnitude of this component is determined by the cosine of the angle between the direction of
this component of the relative velocity, vrel,A-B, and the direction of relative velocity, vrel.

vrel , A B

with

d 2 a2
vrel cos vrel
d2

1/ 2

1
mv 2
2

A B

d 2 a2
d2

The existence of an energy threshold, a, implies that there is a maximum value of a,


above which no reaction occurs.

2
amax
1 a d2

2
d 2 amax
d2

The collisional cross section becomes now

rA rB

2
2
amax
1 a d 2 1 a rA rB

It is smaller than the pure geometric one applied in equation 4.10.

34

35

4.3 Molecularity of reactions


Elementary reactions are those reactions that proceed in a single step, i.e. reactions that
result from a simple collision process (A particle collides with B particle producing a product
molecule). Elementary reactions can be monomolecular, bimolecular or termolecular. Higher
molecularities are not possible because the probability of four or higher number of particles to
meet by chance at one point is infinitely small that it is practically zero.
Bimolecular reactions are easy to understand: Two particles collide with each other. But
what does mean to have a monomolecular reaction. A single particle doesn't react. It needs
another particle to gain energy from in order to be able to overcome the energy barrier. In
photochemical reactions, this second particle is a photon. In normal gas phase reactions, the
second particle can be another reactant particle or an inert particle M.

This issue was addressed by Lindemann who proposed the following mechanism for socalled monomolecular reactions:
A+M
A*

A* + M

-1
2

product

The first step represents the activation of the reactant molecule by collision with an inert
particle M (or another reactant molecule A). The A particle gains energy and becomes excited
(A*). Applying the steady state approximation yields:

36

d [A*]
k1 AM k1 A*M k2 A* 0
dt
k1 AM
A*
k1M k2

k2k1 AM
rate k2 A*
k1 M k2

(4.13)

Normal pressure is considered to be high. Thus, [M] is high and it follows that
k-1[M]>>k2, and a first order kinetics is obtained.

k2k1 A M k2k1
rate k2 A *
A k A
k1 M k1

However, at low pressures, k-1[M]<<k2 and a second order kinetics is observed.

k2k1 AM
2

rate k2 A *
k1 A M or k1 A
k1 M k2

The result is experimentally verified in accordance with Lindemann predictions. An


example is N2O5 NO2 + NO3.
If equation 4.13 is to be written in the form of first-order kinetics, then

rate

k2 k1 A M
k k M
A kobs A
2 1
k 1 M k2 k1 M k2

37

kobs is thus a function of the concentration of the inert substance M (or A itself in case the
required energy to overcome the barrier has been provided by the collision A-A).

ko

kobs

k1/2

[M]

[M]1/2

From the figure above, ko, k1/2 and [M]1/2 can be determined. [M]1/2 is the concentration of
M where kobs equals ko/2 (k1/2).
kobs

k2 k1 M
k1 M k2
ko

k1 / 2

k2 k1
k1

k 2 k1
k2 k1 M 1 / 2

2 k 1
k1 M 1 / 2 k 2

k 1 M 1 / 2 k 2 2 k1 M 1 / 2
k 2 k1 M 1 / 2

ko

k 2 k1
k M 1 / 2 k1
1
M 1 / 2 k1
k1
k1
k1

ko
M 1 / 2

Thus, k1 can be experimentally determined. The steric factor for the process can be
calculated applying equation (4.12).
Termolecular reactions need also some attention. According to the hard sphere model
where particles are treated as if they were rigid billiard balls (section 4.2), the probability of
termolecular collisions is practically zero. This is because the two particles depart as soon as they

38

collide, and the chance of meeting a third particle at the same instant of their collision is
negligibly small. But in real molecules, the two colliding particles stay attached for a short time
which is in the magnitude of a single vibration (10-13s), raising thus the chance of meeting a third
particle.1 In this short time, gas particles move in average a distance of 10-8cm (let's call this
distance ). The cross section for effective collisions is thus a circle with the radius . The
number of collisions is given by:

Z ABC 8 rA rB rC rB 2 3k BT
2

1/ 2

1
1
1 / 2 N A N B N C
1/ 2
CB
AB

At room temperature and 1 atm, ZABC is in the range 1026-1027 cm-3s-1. This is two orders of
magnitude smaller than ZAB. This is the reason why elementary termolecular reactions proceed
much slower than the bimolecular ones.

C
B

A
uneffective

effective

A last comment concerning termolecular reactions: Recombination reactions, as the one


presented below, are not really bimolecular. Since bond energy releases energy, a third particle
must be present in order to dissipate this energy and stabilize the product molecule. Otherwise the
"formed" bond will break down!
H + H H2

H=-BE

H + H + M H2 + M

Hence, termolecular collisions can be considered to consist of two consecutive bimolecular collisions.

39

40

You might also like