21A Collision theory

kr = Ae − E /RT
a Arrhenius expression (21A.1b)
Contents
21A.1 Reactive encounters 881 where A is the ‘pre-exponential factor’ and Ea is the ‘activa-
(a) Collision rates in gases 882 tion energy’. The model is then improved by examining how
Brief illustration 21A.1: Collision density 882 the energy of a collision is distributed over all the bonds in the
(b) The energy requirement 883 reactant molecule. This improvement helps to account for the
Brief illustration 21A.2: The rate constant 884 value of the rate constant kb that appears in the Lindemann the-
(c) The steric requirement 885 ory of unimolecular reactions (Topic 20F).
Brief illustration 21A.3: The steric factor 885
Example 21A.1: Estimating a steric factor 885
21A.2 The RRK model 886
Brief illustration 21A.4: The RRK model 887 21A.1 Reactive encounters
Checklist of concepts 888
Checklist of equations 888 We can anticipate the general form of the expression for kr in
eqn 21A.1a by considering the physical requirements for reac-
tion. We can expect the rate v to be proportional to the rate of
collisions, and therefore to the mean speed of the molecules,
vmean ∝ (T/M)1/2 where M is some combination of the molar
➤ Why do you need to know this material? masses of A and B; we also expect the rate to be proportional to
A major component of chemistry is the study of the their collision cross-section, σ, (Topic 1B) and to the number
mechanisms of chemical reactions. One of the earliest densities NA and NB of A and B:
approaches, which continues to give insight into the
details of mechanisms, is collision theory. v ∝ σ (T /M )1/2 N A N B ∝ σ (T /M )1/2[A][B]

➤ What is the key idea?
According to collision theory, in a bimolecular gas-phase
reaction, a reaction takes place on the collision of reactants
provided their relative kinetic energy exceeds a threshold
value and certain steric requirements are fulfilled.
➤ What do you need to know already?
This Topic draws on the kinetic theory of gases (Topic 1B)
and extends the account of unimolecular reactions (Topic
20F). The latter uses combinatorial arguments like those
described in Topic 15A.
However, a collision will be successful only if the kinetic energy
exceeds a minimum value which we denote E′. This require-
ment suggests that the rate should also be proportional to a
Boltzmann factor of the form e− E′/RT representing the fraction
of collisions with at least the minimum required energy E′.
Therefore,
v ∝ σ (T /M )1/2 e − E ′/RT [A][B]
and we can anticipate, by writing the reaction rate in the form
given in eqn 21A.1, that

kr ∝ σ (T /M )1/2 e − E ′/RT
In this Topic we consider the bimolecular elementary reaction
At this point, we begin to recognize the form of the Arrhenius
A+B→P v = kr[A][B] (21A.1a) equation, eqn 21A.1b, and identify the minimum kinetic energy
E′ with the activation energy Ea of the reaction. This identifica-
where P denotes products. Our aim is to calculate the second- tion, however, should not be regarded as precise, since collision
order rate constant kr and to justify the form of the Arrhenius theory is only a rudimentary model of chemical reactivity.
expression (Topic 20D):
882 21 Reaction dynamics

Not every collision will lead to reaction even if the energy For like molecules µ = 12 mA and at a molar concentration [A]
requirement is satisfied, because the reactants may need to col-
1/2 1/2
lide in a certain relative orientation. This ‘steric requirement’ ⎛ 16kT ⎞ ⎛ 4kT ⎞
Z AA = 12 σ ⎜ N A2 [A]2 = σ ⎜ N A2 [A]2 (21A.3d)
suggests that a further factor, P, should be introduced, and that ⎝ πmA ⎟⎠ ⎝ πmA ⎟⎠

kr ∝ Pσ (T /M )1/2 e − E ′/RT (21A.2) The (blue) factor of 12 is included to avoid double counting of
collisions in this instance. If the collision density is required in
As we shall see in detail below, this expression has the form pre- terms of the pressure of each gas J, then we use [J] = nJ/V = pJ/RT.
dicted by collision theory. It reflects three aspects of a success-
ful collision: Brief illustration 21A.1 Collision density
Minimum
Steric Encounter energy Collision densities may be very large. For example, in nitrogen
requirement "$rate
#$ % requirement
"#%
! at 25 °C and 1.0 bar, when [N2] ≈ 40 mol m− 3, with σ = 0.43 nm2
kr ∝ P σ (T /M ) 1/2
e − E ′/RT and mN2 = 28 .0 2mu the collision density is
1/2
⎛ 4 × (1.381 × 10−23 JK −1) × (298 K) ⎞
(a) Collision rates in gases Z N2 N2 = (4.3 ×10−19 m 2 ) × ⎜
⎝ π × 28.02 × (1.661 ×10−27 kg) ⎟⎠
We have anticipated that the reaction rate, and hence kr, × (6.022 ×1023 mol −1 )2 × (40 mol m −3 )2 = 8.4 ×10034 m −3 s −1
depends on the frequency with which molecules collide. The
collision density, ZAB, is the number of (A,B) collisions in a Even in 1 cm 3 , there are over 8 × 1016 collisions in each
region of the sample in an interval of time divided by the vol- picosecond.
ume of the region and the duration of the interval. The fre- Self-test 21A.1 Calculate the collision density in molecular
quency of collisions of a single molecule in a gas was calculated hydrogen under the same conditions.
in Topic 1B (eqn 1B.11a, z = σvNA). As shown in the following
Answer: Z H2 H2 = 2.0 × 1035 m− 3 s − 1
Justification, that result can be adapted to deduce that

1/2
⎛ 8 kT ⎞
Z AB = σ ⎜ N A2 [A][B] KMT (21A.3a)
⎝ πµ ⎟⎠
Collision density Justification 21A.1 The collision density
It follows from Topic 1B that the collision frequency, z, for a
where σ is the collision cross-section (Fig. 21A.1) single A molecule of mass m A in a gas of other A molecules
is z = σv relNA, where NA is the number density of A molecules
σ = πd 2 d = 12 (dA + dB ) Collision cross-section (21A.3b) and v rel is their relative mean speed. As indicated in Topic 1B,
v rel = 21/2v mean with v mean = (8kT/πm)1/2. For future convenience,
dA and dB are the diameters of A and B, respectively, and µ is the it is sensible to introduce µ = 12 m (for like molecules of mass
reduced mass, m), and then to write v rel = (8kT/πµ)1/2 . This expression also
mA mB applies to the mean relative speed of dissimilar molecules pro-
µ= Reduced mass (21A.3c) vided that µ is interpreted as their reduced mass.
mA + mB
The total collision density is the collision frequency multi-
plied by the number density of A molecules:
dA Z AA = 12 zN A = 12 σ vrelN A2

A The factor of 12 has been introduced to avoid double counting

d of the collisions (so one A molecule colliding with another A
molecule is counted as one collision regardless of their actual
dB B identities). For collisions of A and B molecules present at num-
ber densities NA and NB, the collision density is

Area σ Z AB = σ vrelN AN B

Figure 21A.1 The collision cross-section for two molecules
can be regarded to be the area within which the projectile
molecule (A) must enter around the target molecule (B) in order
for a collision to occur. If the diameters of the two molecules
are dA and dB, the radius of the target area is d = 21 (dA + dB ) and
the cross-section is πd2.
The factor of 12 has been discarded because now we are consid-
ering an A molecule colliding with any of the B molecules as a
collision. The number density of a species J is NJ = NA[J], where
[J] is their molar concentration and NA is Avogadro’s constant.
Equation 21A.3 then follows.
 21A Collision theory  883

1.2
(b) The energy requirement
1 1.7 × 104 pm2, 1.95 eV
According to collision theory, the rate of change in the
number density, NA, of A molecules is the product of the 0.8
collision density and the probability that a collision occurs

σ(ε)/σ
with sufficient energy. The latter condition can be incorpo- 0.6 1.0 × 104 pm2, 0.90 eV
rated by writing the collision cross-section σ as a function
0.4
of the kinetic energy ε of approach of the two colliding spe-
cies, and setting the cross-section, σ (ε), equal to zero if the 0.2
kinetic energy of approach is below a certain threshold value,
εa. Later, we shall identify NAεa as Ea, the (molar) activation 0
0 5 10 15
ε/εa
energy of the reaction. Then, for a collision between A and
B with a specific relative speed of approach vrel (not, at this
Figure 21A.2 The variation of the reactive cross-section with
stage, a mean value),
energy as expressed by eqn 21A.7. The data points are from
experiments on the reaction H + D2 → HD + D (K. Tsukiyama
dN A et al., J. Chem. Phys. 84, 1934 (1986)).
= −σ (ε )vrel N A N B (21A.4a)
dt

or, in terms of molar concentrations,

Justification 21A.2 The collision cross-section
d[A] Consider two colliding molecules A and B with relative
= −σ (ε )vrel N A [A][B] (21A.4b)
dt speed v rel and relative kinetic energy ε = 12 µvrel
2
(Fig. 21A.3).
Intuitively, we expect that a head-on collision between A and
The kinetic energy associated with the relative motion of the B will be most effective in bringing about a chemical reaction.
two particles takes the form ε = 12 µvrel
2
when the centre-of-mass Therefore, v rel,A− B, the magnitude of the relative velocity com-
coordinates are separated from the internal coordinates of each ponent parallel to an axis that contains the vector connecting
particle. Therefore the relative speed is given by vrel = (2 ε /µ )1/2 . the centres of A and B, must be large. From trigonometry and
At this point we recognize that a wide range of approach ener- the definitions of the distances a and d and the angle θ given
gies ε is present in a sample, so we should average the expres- in Fig. 21A.3, it follows that
sion just derived over a Boltzmann distribution of energies f(ε),
1/2
and write ⎛ d 2 − a2 ⎞
vrel , A−B = vrel cosθ = vrel ⎜
⎝ d 2 ⎟⎠
d[A] ⎧ ∞
⎫
dt
= −⎨
⎩ ∫ 0
σ (ε )vrel f (ε )dε ⎬ N A [A][B]
⎭
(21A.5)

d
and hence recognize the rate constant as B
a
vrel, A–B
∞
kr = N A ∫
0
σ (ε )vrel f (ε )dε Rate constant (21A.6) A
θ
vrel
(d 2 – a2)1/2
Now suppose that the reactive collision cross-section is zero
below εa. We show in the following Justification that, above εa,
σ(ε) varies as Figure 21A.3 The parameters used in the calculation of the
dependence of the collision cross-section on the relative
⎛ ε ⎞ kinetic energy of two molecules A and B.
σ (ε )= ⎜ 1 − a ⎟ σ Energy dependence of σ (21A.7)
⎝ ε⎠ We assume that only the kinetic energy associated with the
head-on component of the collision, εA–B, can lead to a chemi-
with the energy-independent σ given by eqn 21A.3b. This form cal reaction. After squaring both sides of this equation and
of the energy-dependence for σ (ε) is broadly consistent with multiplying by 12 µ, it follows that
experimental determinations of the reaction between H and D2
as determined by molecular beam measurements of the kind d 2 − a2
ε A −B = ε ×
described in Topic 21D (Fig. 21A.2). d2
884 21 Reaction dynamics

It follows that
The existence of an energy threshold, εa, for the formation of
products implies that there is a maximum value of a, a max, ∞
⎛ 8 kT ⎞
1/2

above which reaction does not occur. Setting a = a max and ∫ 0

σ (ε )vrel f (ε )dε = σ ⎜
⎝ π µ ⎟⎠
e −εa /kT
εA–B = εa gives
as in eqn 21A.8 (with εa/kT = Ea/RT).
⎛ ε ⎞
2
amax = ⎜1− a ⎟ d 2
⎝ ε⎠

Substitution of σ (ε) for πamax

2
and σ for πd 2 in the equation Equation 21A.8 has the Arrhenius form kr = Ae − E /RT pro- a

above gives eqn 21A.7. Note that the equation can be used only vided the exponential temperature dependence dominates the
when ε > εa. weak square-root temperature dependence of the pre-exponen-
tial factor. It follows that we can identify (within the constraints
of collision theory) the activation energy, Ea, with the mini-
With the energy dependence of the collision cross-section mum kinetic energy along the line of approach that is needed
established, we can evaluate the integral in eqn 21A.6. In the for reaction, and that the pre-exponential factor is a measure of
following Justification we show that the rate at which collisions occur in the gas.
The simplest procedure for calculating kr is to use for σ the
kr = σN A vrel e − E /RT a
Collision theory Rate constant (21A.8) values obtained for non-reactive collisions (for example, typi-
cally those obtained from viscosity measurements) or from
tables of molecular radii. If the collision cross-sections of A
and B are σ A = πdA2 and σ B = πdB2, then an approximate value
Justification 21A.3 The rate constant of the AB cross-section can be estimated from σ = π d2, with
d = 12 (dA + dB ) . That is,
The Maxwell–Boltzmann distribution of molecular speeds is
eqn 1B.4 of Topic 1B:
3/2
σ ≈ 14 (σ 1A/2 + σ B1/2 )2
⎛ µ ⎞
f (v)dv = 4 π ⎜ v 2e − µv /2 kT dv
2

⎝ 2π kT ⎟⎠

(We have replaced M/R by µ/k.) This expression may be writ-

Brief illustration 21A.2 The rate constant
ten in terms of the kinetic energy, ε, by writing ε = 12 µv 2 ; then
dv= dε/(2µε)1/2, when it becomes To estimate the rate constant for the reaction H2 + C2H4 → C2H6
3/2 at 628 K we f irst ca lcu late t he reduced mass using
⎛ µ ⎞ ⎛ 2 ε ⎞ − ε /kT dε m(H2) = 2.016mu and m(C2H4) = 28.05mu. A straightforward cal-
f (v)dv = 4 π ⎜ ⎜⎝ µ ⎟⎠ e
⎝ 2 π kT ⎟⎠ (2 µε )1/2 culation gives µ = 3.123 × 10− 27 kg. It then follows that
3/2
⎛ 1 ⎞
= 2π ⎜ ε 1/2 e − ε /kT dε = f (ε )dε 1/2
⎝ π kT ⎟⎠
1/2
⎛ 8kT ⎞ ⎛ 8 × (1.381 ×10 −23 JK −1) × (628 K) ⎞
⎜⎝ πµ ⎟⎠ =⎜ ⎟⎠ = 2.65 …km s −1
⎝ π × (3.123 ×10 −27 kg))
The integral we need to evaluate is therefore
(2ε /µ )1/2 From Table 1B.1, σ (H 2) = 0.27 nm 2 and σ (C 2H4) = 0.64 nm 2 ,
∞ ! ⎛ 1 ⎞
3/2 ∞
⎛ 2ε ⎞
1/2
giving σ (H 2 ,C 2H4) ≈ 0.44 nm 2 . The activation energy, Table
∫ 0
σ (ε ) vrel f (ε )dε = 2 π ⎜
⎝ π kT ⎟⎠ ∫
0
σ (ε )⎜ ⎟ ε 1/2 e − ε /kT dε
⎝ µ⎠ 20D.1, is large: 180 kJ mol− 1. Therefore,
1/2 ∞
⎛ 8 ⎞ ⎛ 1 ⎞
=⎜
⎝ π µkT ⎟⎠ ⎜⎝ kT ⎟⎠ ∫ 0
ε σ(ε )e − ε /kT dε kr = (4.4 ×10−19 m 2 ) × (2.65…× 103 ms −1) × (6.022 × 1023 mol −1)
× e −(1.800×10 J mol −1 )/(8.3145 J K −1 mol −1 )×(628 K )
5

To proceed, we introduce the approximation for σ(ε) in eqn !####" A #### $

21A.7 and evaluate = 7.04…× 108 m 3 mol −1 s −1 × e −34.4… = 7.5 × 10−7 m 3 mol −1 s −1
σ = 0 for ε <ε a
∞ ! ∞
⎛ ε ⎞
∫0
εσ (ε )e − ε /kT dε = σ ∫ ε ⎜ 1 − a ⎟ e − ε /kT dε
εa ⎝ ε⎠
or 7.5 × 10− 4 dm3 mol− 1 s− 1.
Self-test 21A.2 Evaluate the rate constant for the reaction
⎧ ∞ ∞
⎫
=σ ⎨
⎩ ∫ εa
εe − ε /kT
dε − ∫εa
εae − ε /kT
dε ⎬
⎭
NO + Cl 2 → NOCl + Cl at 298 K from σ (NO) = 0.42 nm 2 and
σ(Cl2) = 0.93 nm2 and data in Table 1B.1.
Integral E.1
! Answer: 2.7 × 10 − 4 dm3 mol− 1 s − 1
= (kT )2 σ e − εa /kT
 21A Collision theory  885

(c) The steric requirement Brief illustration 21A.3 The steric factor
Table 21A.1 compares some values of the pre-exponential fac- It is found experimentally that the pre-exponential factor for the
tor calculated from the collisional data in Table 1B.1 with val- reaction H2 + C2H4 → C2H6 at 628 K is 1.24 × 106 dm3 mol− 3 s − 1.
ues obtained from Arrhenius plots. One of the reactions shows In Brief illustration 21A.2 we calculated the result that can be
fair agreement between theory and experiment, but for others expressed as A = 7.04… × 1011 dm3 mol− 1 s − 1. It follows that the
there are major discrepancies. In some cases the experimental steric factor for this reaction is
values are orders of magnitude smaller than those calculated,
Aexperimental 1.2 4 × 10 6 dm 3 mol −1 s −1
which suggests that the collision energy is not the only crite- P= = ≈ 1.8 ×10 −6
Acalculated 7.0 4…× 10 11 dm 3 mol −1 s −1
rion for reaction and that some other feature, such as the rela-
tive orientation of the colliding species, is important. Moreover, The very small value of P is one reason why catalysts are
one reaction in the table has a pre-exponential factor larger needed to bring this reaction about at a reasonable rate. As a
than theory, which seems to indicate that the reaction occurs general guide, the more complex the reactant molecules, the
more quickly than the particles collide! smaller the value of P.
The disagreement between experiment and theory can be
Self-test 21A.3 It is found for the reaction NO + Cl2 → NOCl + Cl
eliminated by introducing a steric factor, P, and expressing
that A = 4.0 × 109 dm3 mol− 1 s− 1 at 298 K. Estimate the P factor for
the reactive cross-section, σ *, as a multiple of the collision the reaction (see Self–test 21A.2).
cross-section, σ * = Pσ (Fig. 21A.4). Then the rate constant
Answer: 0.019
becomes
1/2
⎛ 8 kT ⎞
kr = PσN A ⎜ e − E /RT
⎝ π µ ⎟⎠
a
(21A.9) An example of a reaction for which it is possible to estimate
the steric factor is K + Br2 → KBr + Br, with the experimental
This expression has the form we anticipated in eqn 21A.2. The value P = 4.8. In this reaction, the distance of approach at which
steric factor is normally found to be several orders of magni- reaction occurs appears to be considerably larger than the
tude smaller than 1. distance needed for deflection of the path of the approaching
molecules in a non-reactive collision. It has been proposed that
the reaction proceeds by a harpoon mechanism. This brilliant
Table 21A.1* Arrhenius parameters for gas-phase reactions name is based on a model of the reaction which pictures the
K atom as approaching a Br2 molecule, and when the two are
A/(dm3 mol− 1 s− 1) Ea/(kJ mol− 1) P
close enough an electron (the harpoon) flips across from K to
Experiment Theory
Br2. In place of two neutral particles there are now two ions, so
2 NOCl → 9.4 × 109 5.9 × 1010 102 0.16 there is a Coulombic attraction between them: this attraction
2 NO + 2 Cl is the line on the harpoon. Under its influence the ions move
2 ClO → Cl2 + O2 6.3 × 107 2.5 × 1010 0 2.5 × 10− 3 together (the line is wound in), the reaction takes place, and
H2 + C2H4 → C2H6 1.24 × 106 7.4 × 1011 180 1.7 × 10− 6 KBr + Br emerge. The harpoon extends the cross-section for the
K + Br2 → KBr + Br 1.0 × 1012 2.1 × 1011 0 4.8 reactive encounter, and the reaction rate is significantly under-
* More values are given in the Resource section. estimated by taking for the collision cross-section the value for
simple mechanical contact between K and Br2.

Deflected reactant
Example 21A.1 Estimating a steric factor
molecule

Estimate the value of P for the harpoon mechanism by calcu-

lating the distance at which it becomes energetically favour-
Products able for the electron to leap from K to Br2. Take the sum of the
Area σ* radii of the reactants (treating them as spherical) to be 400 pm.
Area σ Method Begin by identif ying all the contributions to
the energy of interaction between the colliding species.
Figure 21A.4 The collision cross-section is the target area There are three contributions to the energy of the process
that results in simple deflection of the projectile molecule; the K + Br2 → K + + Br2− . The first is the ionization energy, I, of K.
reactive cross-section is the corresponding area for chemical The second is the electron affinity, Eea, of Br2. The third is the
change to occur on collision.
886 21 Reaction dynamics

of the model is that although a molecule might have enough

Coulombic interaction energy between the ions when they
have been formed: when their separation is R, this energy is energy to react, that energy is distributed over all the modes
− e2/4πε 0R. The electron flips across when the sum of these of motion of the molecule, and reaction will occur only when
three contributions changes from positive to negative (that is, enough of that energy has migrated into a particular location
when the sum is zero) and becomes energetically favourable. (such as a particular bond) in the molecule. This distribution
effect leads to a P factor of the form
Answer The net change in energy when the transfer occurs at
a separation R is s −1
⎛ E* ⎞
e2
P = ⎜1 − ⎟ RRK theory (21A.10a)
E = I − Eea − ⎝ E⎠
4πε 0 R
where s is the number of modes of motion over which the
The ionization energy I is larger than Eea, so E becomes nega-
energy E may be dissipated and E* is the energy required for
tive only when R has decreased to less than some critical value
the bond of interest to break. The resulting Kassel form of the
R* given by
unimolecular rate constant for the decay of A* to products is
e2
R=
4 πε 0 (I − Eea ) s −1
⎛ E* ⎞
When the particles are at this separation, the harpoon shoots kb (E )= ⎜ 1 − ⎟ kb for E ≥ E * Kassel form (21A.10b)
⎝ E⎠
across from K to Br2, so we can identify the reactive cross-sec-
tion as σ * = πR*2. This value of σ * implies that the steric factor
is where kb is the rate constant used in the original Lindemann
theory for the decomposition of the activated intermediate
2
σ * R *2 ⎛ e2 ⎞ (eqn 20F.8 of Topic 20F).
P= = 2 =⎜ ⎟
σ d ⎝ 4π ε 0 d ( I − E ea ⎠
)

where d = R(K) + R(Br2), the sum of the radii of the spherical
reactants. With I = 420 kJ mol− 1 (corresponding to 0.70 aJ),
Eea ≈ 250 kJ mol− 1 (corresponding to 0.42 aJ), and d = 400 pm,
we find P = 4.2, in good agreement with the experimental
value (4.8).
Self-test 21A.4 Estimate the value of P for the harpoon reac-
tion between Na and Cl2 for which d ≈ 350 pm; take Eea ≈ 230 kJ
mol− 1.
Answer: 2.2
Example 21A.1 illustrates two points about steric factors.
First, the concept of a steric factor is not wholly useless because
in some cases its numerical value can be estimated. Second, and
more pessimistically, most reactions are much more complex
than K + Br2, and we cannot expect to obtain P so easily.
Justification 21A.4 The RRK model of unimolecular
reactions
To set up the RRK model, we suppose that a molecule con-
sists of s identical harmonic oscillators, each of which has
frequency ν. In practice, of course, the vibrational modes of
a molecule have different frequencies, but assuming that they
are all the same is a reasonable first approximation. Next, we
suppose that the vibrations are excited to a total energy E = nhν
and then set out to calculate the number of ways N in which
the energy can be distributed over the oscillators.
We can represent the n quanta as follows:
,,,,,,,,,,,,,,,,,,,,,,,,,,
,,,,,,…,,,
These quanta must be put in s containers (the s oscillators),
which can be represented by inserting s − 1 walls, denoted by  .
One such distribution is

|
,,,,,,,,| ||,,,,,,,,,,,,,,,
| | |||
21A.2 The RRK model |
,,,,,,,,, …,,,
|

The steric factor P can also be estimated for unimolecular
gas-phase reactions (Topic 20F), by a calculation based on
the Rice–Ramsperger–Kassel model (RRK model), which
was proposed in 1926 by O.K. Rice and H.C. Ramsperger and
almost simultaneously by L.S. Kassel. The model has been elab-
orated, largely by R.A. Marcus, into the ‘RRKM model’. Here we
outline Kassel’s original approach to the RRK model; the details
are set out in the following Justification. The essential feature
The total number of arrangements of each quantum and wall
(of which there are n + s − 1 in all) is (n + s − 1)! where, as usual,
x! = x(x − 1)…1. However, the n! arrangements of the n quanta
are indistinguishable, as are the (s − 1)! arrangements of the
s − 1 walls. Therefore, to find N we must divide (n + s − 1)! by
these two factorials. It follows that
(n + s − 1)!
N=
n !(s − 1)!
 21A Collision theory  887

The distribution of the energy throughout the molecule !### −1 "

s# factors
#### $ s −1
means that it is too sparsely spread over all the modes for any (n − n*)(n − n*)…(n − n*) ⎛ n − n * ⎞
P= =⎜
particular bond to be sufficiently highly excited to undergo n)(
(% #n& )…#(' n) ⎝ n ⎟⎠
dissociation. We suppose that a bond will break only if it is s−1 factors
excited to at least an energy E* = n*hν. Therefore, we isolate one
critical oscillator as the one that undergoes dissociation if it An alternative derivation of this expression for P is developed
has at least n* of the quanta, leaving up to n − n* quanta to be in Problem 21A.7. Because the energy of the excited molecule
accommodated in the remaining s − 1 oscillators (and there- is E = nhν and the critical energy is E* = n*hν, this expression
fore with s − 2 walls in the partition in place of the s − 1 walls we may be written
used above). For example, consider 28 quanta distributed over s −1
⎛ E* ⎞
six oscillators, with excitation by at least six quanta required P = ⎜1 − ⎟
⎝ E⎠
for dissociation. Then all the following partitions will result in
dissociation: as in eqn 21A.10.
| | |
,,,,,,,,,,,,,,,,,,,,,,,||,,,,,
,,,,,,,|,,,,,,,,,,,,,,,
| | ,||,,,,,
,,,,,,,,|,,,,,,,,,,,,,,,
| | ||,,,,, The energy dependence of the rate constant given by eqn
21A.10b is shown in Fig. 21A.5 for various values of s. We see
" " " " "
that the rate constant is smaller at a given excitation energy if
(The leftmost partition is the critical oscillator.) However, s is large, as it takes longer for the excitation energy to migrate
these partitions are equivalent to through all the oscillators of a large molecule and accumulate in
the critical mode. As E becomes very large, however, the term
,,,,,, |,,,,,,,,,,,,,,,,,
| | ||,,,,, in parentheses approaches 1, and kb(E) becomes independent
,,,,,, ,|,,,,,,,,,,,,,,,
| | ,||,,,,, of the energy and the number of oscillators in the molecule, as
,,,,,, ,,|,,,,,,,,,,,,,,,
| | ||,,,,,
there is now enough energy to accumulate immediately in the
critical mode regardless of the size of the molecule.
" " " " "

and we see that we have the problem of permuting 28 − 6 = 22

(in general, n − n*) quanta and five (in general, s − 1) walls, 1
and therefore a total of 27 (in general, n − n* + s − 1 objects).
Therefore, the calculation is exactly like the one above for N, 0.8
Rate constant, kb(E)/kb

except that we have to find the number of distinguishable per-

mutations of n − n* quanta in s containers (and therefore s − 1 0.6
5
walls). The number N* is therefore obtained from the expres- 10
sion for N by replacing n by n − n* and is
0.4
20
(n − n * + s −1)!
N*= 0.2
(n − n*)!(s −1)!

From the preceding discussion we conclude that the probabil- 0

0 10 20 30 40 50
ity that one specific oscillator will have undergone sufficient
Relative energy, E/E*
excitation to dissociate is the ratio N*/N, which is
Figure 21A.5 The energy dependence of the rate constant
N * n !(n − n* + s −1)!
P= = given by eqn 21A.10b for three values of s.
N (n − n*)!(n + s −1)!

This equation is still awkward to use, even when written out in

terms of its factors:
Brief illustration 21A.4 The RRK model
n(n −1)(n − 2)…1 (n − n * + s −1)(n − n * + s − 2)…1
P= ×
(n − n*)(n − n * −1)…1 (n + s −1)(n + s − 2)…1 In Brief illustration 21A.3 we calculated a value of P = 1.8 × 10− 6
(n − n * + s −1)(n − n * + s − 2)…(n − n * +1) for the reaction H2 + C2H4 → C2H6. Although this is not a uni-
=
(n + s −1)(n + s − 2)…(n + 2)(n + 1) molecular process, it is interesting to analyse it on the basis
of the RRK theory because in some sense the collision energy
However, because s − 1 is small (in the sense s − 1 ≪ n − n*), we must accumulate in a region where bonds are broken and
can approximate this expression by
888 21 Reaction dynamics

formed. Thus, C2H4 has six atoms and therefore s = 12 vibra- 70 per cent of the energy of a typical collision. If all eight atoms
tional modes. We can estimate the ratio E*/E by solving are taken to be involved in sharing the energy of the collision,
the ratio works out as 0.54.
11
⎛ E* ⎞ E*
⎜⎝ 1 − E ⎟⎠ = 1.8 × 10 = 1 − (1.8 × 10−6 )1/11 = 0.7 0 Self-test 21A.5 Apply the same analysis to the reaction in
−6
or
E
Self-test 21A.3, where it is found that P = 0.019 for NO + Cl2 →
NOCl + Cl. Take the number of atoms in the complex to be 4,
This result suggests in one interpretation that the energy
so s = 6.
needed to proceed in the reaction (identified here with the
energy to break the carbon-carbon bond in C2H4) is typically Answer: 0.55

Checklist of concepts
☐ 1. In collision theory, it is supposed that the rate is pro-
portional to the collision frequency, a steric factor, and
the fraction of collisions that occur with at least the
kinetic energy Ea along their lines of centres.
☐ 2. The collision density is the number of collisions in
a region of the sample in an interval of time divided
by the volume of the region and the duration of the
interval.
☐ 3. The activation energy is the minimum kinetic energy
along the line of approach of reactant molecules that is
required for reaction.
☐ 4. The steric factor is an adjustment that takes into
account the orientational requirements for a successful
collision.
☐ 5. For unimolecular reactions, the steric factor can be
computed by using the RRK model.

Checklist of equations
Property Equation Comment Equation number

Collision density Z AB =σ (8 kT /πµ )1/2 N A2 [A][B] Unlike molecules, KMT (kinetic molecular theory) 21A.3a

Z AA =σ (4 kT /πmA )1/2 N A2 [A]2 Like molecules, KMT 21A.3d

Energy-dependence of σ σ (ε) = (1 − εa/ε)σ ε ≥εa, 0 otherwise 21A.7

Rate constant kr = PσN A (8 kT /πµ )1/2 e − Ea /RT KMT, collision theory 21A.9

Steric factor P = (1 − E*/E)s− 1 RRK theory 21A.10a

21B Diffusion-controlled reactions

less frequent than in a gas. However, because a molecule also

Contents migrates only slowly away from a location, two reactant mol-
ecules that encounter each other stay near each other for much
21B.1 Reaction in solution 889 longer than in a gas. This lingering of one molecule near another
(a) Classes of reaction 889 on account of the hindering presence of solvent molecules is
(b) Diffusion and reaction 890 called the cage effect. Such an encounter pair may accumulate
Brief illustration 21B.1: Diffusion control 1 890 enough energy to react even though it does not have enough
Brief illustration 21B.2: Diffusion control 2 891 energy to do so when it first forms. The activation energy of a
21B.2 The material-balance equation 891 reaction is a much more complicated quantity in solution than
(a) The formulation of the equation 891 in a gas because the encounter pair is surrounded by solvent
(b) Solutions of the equation 892 and we need to consider the energy of the entire local assembly
Brief illustration 21B.3: Reaction with diffusion 892 of reactant and solvent molecules.
Checklist of concepts 892
Checklist of equations 893
(a) Classes of reaction
The complicated overall process can be divided into simpler
parts by setting up a simple kinetic scheme. We suppose that
➤ Why do you need to know this material? the rate of formation of an encounter pair AB is first order in
Most chemical reactions take place in solution and for a each of the reactants A and B:
thorough grasp of chemistry it is important to understand
what controls their rates and how those rates can be modified. A + B → AB v = kd [A][B]

➤ What is the key idea?
There are two limiting types of chemical reaction mechanism
in solution: diffusion control and activation control.
➤ What do you need to know already?
This Topic makes use of the steady-state approximation
(Topic 20E) and draws on the formulation and solution of
the diffusion equation (Topic 19C). At one point it uses the
Stokes–Einstein relation (Topic 19B).
As we shall see, kd (where the d signifies diffusion) is deter-
mined by the diffusional characteristics of A and B. The
encounter pair can break up without reaction or it can go on to
form products P. If we suppose that both processes are pseudo-
first-order reactions (with the solvent perhaps playing a role),
then we can write
AB → A + B v = kd′[AB]
AB → P v = ka [AB]

The concentration of AB can now be found from the equation

for the net rate of change of concentration of AB:
To consider reactions in solution we have to imagine processes
that are entirely different from those in gases. No longer are d[AB]
= kd [A][B]− kd′[AB]− ka [AB]= 0
there collisions of molecules hurtling through space; now there dt
is the jostling of one molecule through a dense but mobile col- where we have applied the steady-state approximation (Topic
lection of molecules making up the fluid environment. 20E). This expression solves to

kd [A][B]
[AB]=
ka + kd′
21B.1 Reactions in solution
The rate of formation of products is therefore
Encounters between reactants in solution occur in a very dif-
ferent manner from encounters in gases. The encounters d[P] ka kd
= ka [AB]= kr [A][B] kr = (21B.1)
of reactant molecules dissolved in solvent are considerably dt ka + kd′
890 21 Reaction dynamics

Two limits can now be distinguished. If the rate of separation Table 21B.1* Arrhenius parameters for solvolysis reactions in
of the unreacted encounter pair is much slower than the rate solution
at which it forms products, then kd′ ≪ ka and the effective rate
Solvent A/(dm3 mol− 1 s− 1) Ea/(kJ mol− 1)
constant is
(CH3)3CCl Water 7.1 × 1016 100
kk Ethanol 3.0 × 1013 112
kr ≈ a d = kd Diffusion-controlled limit (21B.2a)
ka Chloroform 1.4 × 104 45
CH3CH2Br Ethanol 4.3 × 1011 90
In this diffusion-controlled limit, the rate of reaction is gov-
* More values are given in the Resource section.
erned by the rate at which the reactant molecules diffuse
through the solvent. Because the combination of radicals
involves very little activation energy, radical and atom recombi-
nation reactions are often diffusion-controlled. An activation- Justification 21B.1 Solution of the radial diffusion
controlled reaction arises when a substantial activation energy equation
is involved in the reaction AB → P. Then ka ≪ kd′ and
The general form of the diffusion equation (Topic 19A) corres-
ka kd ponding to motion in three dimensions is DB∇ 2[B](r,t) =
kr ≈ = ka K Activation-controlled limit (21B.2b) ∂[B](r,t)/∂t; therefore, the concentration of B when the sys-
kd′
tem has reached a steady state (∂[B](r,t)]/∂t = 0) satisfies ∇ 2[B]
(r) = 0, with the concentration of B now depending only on
where K is the equilibrium constant for A + B ⇌ AB. In this
location not time. For a spherically symmetrical system, ∇ 2
limit, the reaction proceeds at the rate at which energy accu- can be replaced by radial derivatives alone (see Table 7B.1), so
mulates in the encounter pair from the surrounding solvent. the equation satisfied by [B](r), as [B](r) can now be written, is
Some experimental data are given in Table 21B.1.
d 2 [B](r ) 2 d[B](r )
+ =0
dr 2 r dr
(b) Diffusion and reaction
The rate of a diffusion-controlled reaction is calculated by con- The general solution of this equation is
sidering the rate at which the reactants diffuse together. As
b
shown in the following Justification, the rate constant for a reac- [B](r ) = a +
r
tion in which the two molecules react if they come within a dis-
tance R* of one another is as may be verified by substitution. We need two boundary
conditions to pin down the values of the two constants (a and
kd = 4 πR *DN A (21B.3) b). One condition is that [B](r) has its bulk value [B] as r → ∞.
The second condition is that the concentration of B is zero at
where D is the sum of the diffusion coefficients of the two reac- r = R*, the distance at which reaction occurs. It follows that
tant species in the solution. a = [B] and b = − R*[B], and hence that (for r ≥R*)

⎛ R* ⎞
[B](r ) = ⎜ 1 − ⎟ [B]
Brief illustration 21B.1 Diffusion control 1 ⎝ r ⎠

The order of magnitude of R* is 10 − 7 m (100 nm) and that of D Figure 21B.1 illustrates the variation of concentration
for a species in water is 10 − 9 m 2 s − 1. It follows from eqn 21B.3 expressed by this equation.
that The rate of reaction is the (molar) flux, J, of the reactant B
towards A multiplied by the area of the spherical surface of
kd ≈ 4 π × (10−7 m) × (10−9 m 2 s −1) × (6 .022 × 1023 ) ≈ 109 m 3 mol −1s −1
radius R* through which B must pass:
which corresponds to 1012 dm3 mol− 1 s − 1. An indication that a
reaction is diffusion-controlled is therefore that its rate con- Rate of reaction = 4 πR*2 J
stant is of the order of 1012 dm3 mol− 1 s − 1.
From Fick’s first law (eqn 19C.3 of Topic 19C, J = − D∂[J]/∂x),
Self-test 21B.1 Estimate the rate constant for a diffusion- the flux of B towards A is proportional to the concentration
controlled reaction in benzene (D ≈ 2 × 10 − 9 m 2 s − 1), taking gradient, so at a radius R*:
R* ≈ 100 nm.
⎛ d[B](r ) ⎞ ⎛ 1⎞ D [B]
Answer: 1.5 × 1012 dm3 mol− 1 s − 1 J = DB ⎜ = − DB[B]R* ⎜ − 2 ⎟ = B
⎝ dr ⎟⎠ r =R* ⎝ r ⎠ r = R* R*
 21B Diffusion-controlled reactions  891

1
Brief illustration 21B.2 Diffusion control 2
0.8 The rate constant for the recombination of I atoms in hexane
Concentration, [B](r)/[B]

at 298 K, when the viscosity of the solvent is 0.326 cP (with

0.6 1 P = 10− 1 kg m− 1 s − 1) is
R*
0.4 8 × (8.3145 J K −1 mol −1) × (298 K)
kd = = 2.0 ×10 7 m 3 mol −1 s −1
3 × (3.26 ×10 −4 kg m −1 s −1)
0.2
where we have used 1 J = 1 kg m 2 s − 2 . Because 1 m 3 = 103 dm 3,
this result corresponds to 2.0 × 1010 dm3 mol− 1 s − 1. The experi-
0
0 2 4 6 8 10 mental value is 1.3 × 1010 dm 3 mol− 1 s − 1, so the agreement is
Radius, r/R* very good considering the approximations involved.

Figure 21B.1 The concentration profile for reaction in Self-test 21B.2 Evaluate a typical rate constant for a reaction
solution when a molecule B diffuses towards another taking place in ethanol at 20 °C, for which the viscosity is
reactant molecule and reacts if it reaches R*. 1.06 cP.
Answer: 6.1 × 109 dm3 mol− 1 s − 1
(A sign change has been introduced because we are interested
in the flux towards decreasing values of r.) It follows that

Rate of reaction = 4πR*DB[B] 21B.2 The material-balance equation

The rate of the diffusion-controlled reaction is equal to the
average f low of B molecules to all the A molecules in the
sample. If the bulk concentration of A is [A], the number of
A molecules in the sample of volume V is NA[A]V; the global
flow of all B to all A is therefore 4πR*DBNA[A][B]V. Because it
is unrealistic to suppose that all A molecules are stationary;
we replace DB by the sum of the diffusion coefficients of the
two species and write D = DA + DB. Then the rate of change of
concentration of AB is
d[AB]
= 4πR* DN A[A][B]
dt
Hence, the diffusion-controlled rate constant is as given in
eqn 21B.3.
The diffusion of reactants plays an important role in many
chemical processes, such as the diffusion of O2 molecules
into red blood corpuscles and the diffusion of a gas towards
a catalyst. We can catch a glimpse of the kinds of calculations
involved by considering the diffusion equation (Topic 19C)
generalized to take into account the possibility that the diffus-
ing, convecting molecules are also reacting.
(a) The formulation of the equation
Consider a small volume element in a chemical reactor (or
a biological cell). The net rate at which J molecules enter the
region by diffusion and convection is given by eqn 19C.10 of
Topic 19C, which we repeat here:

We can take eqn 21B.3 further by incorporating the Stokes– ∂[J] ∂2[J] ∂[J]
= D 2 −v Diffusion equation (21B.5)
Einstein equation (eqn 19B.19 of Topic 19B, DJ = kT/6πηRJ) ∂t ∂x ∂x
relating the diffusion constant and the hydrodynamic radius RA
and RB of each molecule in a medium of viscosity η. As this where v is the velocity of the convective flow of J and [J] in gen-
relation is approximate, little extra error is introduced if we eral depends on both position and time. The net rate of change
write RA = RB = 12 R*, which leads to of molar concentration due to chemical reaction is
∂[J]
8RT = − kr[J]
kd = Diffusion-controlled rate constant (21B.4) ∂t
3η
if we suppose that J disappears by a pseudofirst-order reaction.
(The R in this equation is the gas constant.) The radii have can- Therefore, the overall rate of change of the concentration of J is
celled because, although the diffusion constants are smaller
when the radii are large, the reactive collision radius is larger Spread
Change
due to
and the particles need travel a shorter distance to meet. In this non-uniform due to Loss
distribution convection due to Material-
approximation, the rate constant is independent of the identi- ! ! reaction balance (21B.6)
∂[J] ∂2 [J] ∂[J] !
ties of the reactants, and depends only on the temperature and =D − v − k equation
r [J]
the viscosity of the solvent. ∂t ∂x 2 ∂x
892 21 Reaction dynamics

Equation 21B.6 is called the material-balance equation. If the An example of a solution of the diffusion equation in the
rate constant is large, then [J] will decline rapidly. However, if absence of reaction is that given in Topic 19C for a system in
the diffusion constant is large, then the decline can be replen- which initially a layer of n0NA molecules is spread over a plane
ished as J diffuses rapidly into the region. The convection term, of area A:
which may represent the effects of stirring, can sweep material
n0 e − x /4 Dt
2

either into or out of the region according to the signs of v and [J]= (21B.9)
A(πDt )1/2
the concentration gradient ∂[J]/∂x.

When this expression is substituted into eqn 21B.8, we obtain the

(b) Solutions of the equation concentration of J as it diffuses away from its initial surface layer
The material-balance equation is a second-order partial differ- and undergoes reaction in the overlying solution (Fig. 21B.2).
ential equation and is far from easy to solve in general. Some
idea of how it is solved can be obtained by considering the Brief illustration 21B.3 Reaction with diffusion
special case in which there is no convective motion (as in an
unstirred reaction vessel): Suppose 1.0 g of iodine (3.9 mmol I2) is spread over a surface of
area 5.0 cm 2 under a column of hexane (D = 4.1 × 10 − 9 m 2 s − 1).
∂[J] ∂2[J] As it diffuses upwards it reacts with a pseudofirst-order rate
= D 2 − kr[J] (21B.7)
∂t ∂x constant k r = 4.0 × 10− 5 s− 1. By substituting these values into

n0 e − x /4 Dt −krt
2

As may be verified by substitution (Problem 21B.1), if the solu- [J]* =

tion of this equation in the absence of reaction (that is, for A(πDt )1/2
kr = 0) is [J], then the solution [J]* in the presence of reaction
we can construct the following table:
(kr > 0) is
[J]*/(mmol dm− 3) x
[J]* =[J]e −k t r Diffusion with reaction (21B.8) T 1 mm 5 mm 1 cm
100 s 3.72 0 0
1000 s 1.96 0.45 0.005
10 000 s 0.46 0.40 0.25
Concentration of reactant, [J]

Dt = 0.05
Self-test 21B.3 What is the value of [J] at 15 000 s at the same
three locations?
Answer: 0.31, 0.28, 0.21 mmol dm− 3
Dt = 0.10

Even this relatively simple example has led to an equation

Distance from plane, x
Figure 21B.2 The concentration profiles for a diffusing,
reacting system (for example, a column of solution) in which
one reactant is initially in a layer at x = 0. In the absence of
reaction (grey lines), the concentration profiles are the same as
in Fig. 19C.3.
that is difficult to solve, and only in some special cases can
the full material balance equation be solved analytically. Most
modern work on reactor design and cell kinetics uses numeri-
cal methods to solve the equation, and detailed solutions for
realistic environments, such as vessels of different shapes
(which influence the boundary conditions on the solutions)
and with a variety of inhomogeneously distributed reactants,
can be obtained reasonably easily.

Checklist of concepts
☐ 1. A reaction in solution may be diffusion-controlled if ☐ 2. The rate of an activation-controlled reaction is con-
its rate is controlled by the rate at which reactant mol- trolled by the rate at which the encounter pair accumu-
ecules encounter each other in solution. lates sufficient energy.
 21B Diffusion-controlled reactions  893

☐ 3. The material-balance equation relates the overall rate ☐ 4. The cage effect, the lingering of one reactant molecule
of change of the concentration of a species to its rates of near another due to the hindering presence of solvent
diffusion, convection and reaction. molecules, results in the formation of an encounter
pair of reactant molecules.

Checklist of equations
Property Equation Comment Equation number

Diffusion-controlled limit kr = kd v = kd[A][B] for the encounter rate 21B.2a

Activation-controlled limit kr = kdK K for A + B ⇌ AB, ka for the decomposition of AB 21B.2b

Diffusion-controlled rate constant kd = 4πR*DNA D = DA + DB 21B.3

kd = 8RT/3η Assumes Stokes–Einstein relation 21B.4

Material-balance equation ∂[J]/ ∂t First-order reaction 21B.6

= D∂2[J]/ ∂x 2
−v∂[J]/ ∂x − kr[J]