X: x, y or z

! / 1st_Year / HeatMassTransfer / 2_Diffusion

Si :
Mass Generation of

( ∂𝑥 ∂𝑧 )
∂𝑐𝑖 ∂𝑁𝑖,𝑥 ∂𝑁𝑖,𝑦 ∂𝑁𝑖,𝑧
2. Diffusion Balance =− + + + 𝑆𝑖 i (could be r
∂𝑡 ∂𝑦
(Cartesian) for rate of
reaction)

Equations Mass Normally

= −(
∂𝑧 )
∂𝑐𝑖 1 ∂ (𝑟𝑁𝑖 ) 1 ∂𝑁𝑖 ∂𝑁𝑖
Balance + 2 + + 𝑆𝑖 ignore z and
(Cylindrical)
∂𝑡 𝑟 ∂𝑟 𝑟 ∂𝜃 θ

Name Equation Notes

1 ∂ (𝑟 𝑁𝑖 )
Mass 2 Normally

( 𝑟2 𝑟2 sin2 𝜃 ∂𝜙 )
∂𝑐𝑖 1 ∂(𝑁𝑖 sin 𝜃) 1 ∂𝑁𝑖
Ci : Balance =− + 2 + + 𝑆𝑖 ignore θ and
∂𝑡 ∂𝑟 𝑟 sin 𝜃 ∂𝜃
Concentration (Spherical) φ
Molar 𝑛
∑𝑖=1 𝑐𝑖 𝑣𝑖 of i
Average 𝑣𝐴 = 𝑛
Velocity
∑𝑖=1 𝑐𝑖
Fick’s Law
vi : velocity of
for 𝑑𝑥1
i 𝑁1 = 𝑥1 (𝑁1 + 𝑁2 ) − 𝐷𝑐𝑇
Absolute 𝑑𝑧
Flux
Absolute
𝑁𝑖 = 𝑐𝑖 𝑣𝑖 Ni : Molar flux
Flux

The Diffusion Equation

𝐽𝑖 = 𝑐𝑖 (𝑣𝑖 − 𝑣𝐴 )
Ji : Relative Mass transport is defined as the movement of material due to a difference in concentration of that
Diffusive molar flux material. This can be done by diffusion and convective mass transfer. Unfortunately, it is not a parallel of
Flux (Diffusive heat transfer as now we have multiple species with varying concentrations (on a molar basis). The sum of
𝑛 their concentrations is the equivalent of the total concentration:
flux)
∑
𝐽𝑖 = 0
𝑖=1
𝑛

∑ 𝑖
𝑐𝑇 = 𝑐
𝑖=1
CT : Total
concentration
Concentration does differ with position (especially in a small volume dV) and time (when not at steady
Di,j : state).
Molecular
Fick’s Law We also need to consider the velocity of each species which is normally taken as an average given by:
diffusivity of i
for 𝑑𝑥
𝐽𝑖 = −𝑐𝑇 𝐷𝑖,𝑗 𝑖 through j
Diffusive 𝑑𝑋 𝑛
∑𝑖=1 𝑐𝑖 𝑣𝑖
Flux
xi : Molar
𝑣𝐴 = 𝑛
∑𝑖=1 𝑐𝑖
composition
of i
We also use molar compositions when discussion diffusion:
 𝑐𝑖 𝑑𝑥1 𝑑𝑥2
𝑥𝑖 = −𝑐𝑇 𝐷1,2 = 𝑐𝑇 𝐷2,1
𝑐𝑇 𝑑𝑥 𝑑𝑥

There is such a thing as the molar flux which is defined as the number of moles travelling in a unit area for
a unit time: 𝑥1 = 1 − 𝑥2 → 𝑑𝑥1 = −𝑑𝑥2

𝑛𝑖 𝑐𝑖 𝑑𝑉 𝑐𝑖 𝐴𝑑𝑠
𝑁𝑖 = = = = 𝑐𝑖 𝑣𝑖 ∴ 𝐷1,2 = 𝐷2,1
𝐴𝑑𝑡 𝐴𝑑𝑡 𝐴𝑑𝑡
Material Balance
This is known as the absolute flux and is based on a stationary coordinate system. However, this requires
tracking an individual species which can be very difficult. However, it is possible to measure it relative to the Again, taking an infinitesimally small volume:
average velocity:

𝐽𝑖 = 𝑐𝑖 (𝑣𝑖 − 𝑣𝐴 )

This is known as the diffusive flux and it uses the relative velocity rather than the actual velocity. Because
of this, the sum of all the diffusive flux is 0:

𝑛
𝐽𝑖 = 𝑐1 (𝑣1 − 𝑣𝐴 ) + ⋯ + 𝑐𝑛 (𝑣𝑛 − 𝑣𝐴 )
∑
𝑖=1
𝑛 𝑛
1. [ENTER DESCRIPTION]
∑ ∑
= 𝑐𝑖 𝑣𝑖 − 𝑐𝑖 𝑣𝐴
𝑖=1 𝑖=1 The material balance (mirroring the energy balance):
𝑛 𝑛

∑ ∑
= 𝑐𝑖 𝑣𝑖 − 𝑣𝐴 𝑐𝑖 Rate of accumulation = Rate of material in − Rate of material out + Rate of generation
𝑖=1 𝑖=1
𝑛 𝑛 𝑛
∑𝑖=1 𝑐𝑖 ∑𝑖=1 𝑐𝑖 𝑣𝑖 Here we deal with absolute fluxes where the rate of material in is:
= ∑ 𝑐𝑖 𝑣𝑖 − 𝑛
𝑖=1
∑𝑖=1 𝑐𝑖
𝑛 𝑛 𝑁𝑖,𝑥 𝑑𝑦𝑑𝑧𝑑𝑡
∑ ∑
= 𝑐𝑖 𝑣𝑖 − 𝑐𝑖 𝑣𝑖
𝑖=1 𝑖=1
=0 And the rate of material out is:

There exists an empirical law for the diffusive flux termed Fick’s law: 𝑁𝑖,𝑥+𝑑𝑥 𝑑𝑦𝑑𝑧𝑑𝑡

𝑑𝑥𝑖 The generation term is:

𝐽𝑖 = −𝑐𝑇 𝐷𝑖,𝑗
𝑑𝑥
𝑆𝑖 𝑑𝑥𝑑𝑦𝑑𝑧𝑑𝑡
Where D i,j is the diffusivity of i relative to j. Incidentally, for 2 species:

And the accumulation term is:

𝐽1 + 𝐽2 = 0 → 𝐽1 = −𝐽2
 diffusion:

(𝑐𝑡+𝑑𝑡 − 𝑐𝑇 ) 𝑑𝑥𝑑𝑦𝑑𝑧

Altogether:

𝑑𝑐𝑖 𝑑𝑥𝑑𝑦𝑑𝑧 = −𝑑𝑁𝑖,𝑥 𝑑𝑦𝑑𝑧𝑑𝑡 − 𝑑𝑁𝑖,𝑦 𝑑𝑥𝑑𝑧𝑑𝑡 − 𝑑𝑁𝑖,𝑧 𝑑𝑥𝑑𝑦𝑑𝑡 + 𝑆𝑖 𝑑𝑥𝑑𝑦𝑑𝑧𝑑𝑡

Dividing by dx, dy, dz and dt:

( ∂𝑥 ∂𝑧 )
∂𝑐𝑖 ∂𝑁𝑖,𝑥 ∂𝑁𝑖,𝑦 ∂𝑁𝑖,𝑧
=− + + + 𝑆𝑖
∂𝑡 ∂𝑦

This is a parallel to the energy balance, however, we do not have an empirical law for the absolute flux. It is
also only for Cartesian coordinates, for the other systems:
2. [ENTER DESCRIPTION]

= −(
∂𝑧 )
∂𝑐𝑖 1 ∂ (𝑟𝑁𝑖 ) 1 ∂𝑁𝑖 ∂𝑁𝑖
+ 2 + + 𝑆𝑖
∂𝑡 𝑟 ∂𝑟 𝑟 ∂𝜃 𝑑𝑥𝐴
𝑁𝐴 = 𝑥𝐴 (𝑁𝐴 + 𝑁𝐵 ) − 𝑐𝑇 𝐷 ; 𝑁𝐵 = 0
𝑑𝑧

1 ∂ (𝑟 𝑁𝑖 )
2

( 𝑟 sin 𝜃 ∂𝜙 )
∂𝑐𝑖 1 ∂(𝑁𝑖 sin 𝜃) 1 ∂𝑁𝑖
=− + 2 + + 𝑆𝑖 𝑑𝑥𝐴
∂𝑡 𝑟 2 ∂𝑟 𝑟 sin 𝜃 ∂𝜃 2 2
∴ 𝑁𝐴 = 𝑥𝐴 𝑁𝐴 − 𝑐𝑇 𝐷
𝑑𝑧

However, if we only consider a binary system in the z direction:

𝑑𝑥𝐴
𝑁𝐴 (1 − 𝑥𝐴 ) = −𝑐𝑇 𝐷
𝑑𝑧
𝑁1 = 𝑐1 𝑣1 = 𝑐1 (𝑣1 − 𝑣𝐴 ) + 𝑐1 𝑣𝐴 = 𝐽1 + 𝑐1 [
𝑐1 + 𝑐2 ]
𝑐1 𝑣1 + 𝑐2 𝑣2

−𝑐𝑇 𝐷 𝑑𝑥𝐴
𝑁𝐴 =
𝑑𝑥1 1 − 𝑥𝐴 𝑑𝑧
∴ 𝑁1 = −𝑐𝑇 𝐷 + 𝑥1 (𝑁1 + 𝑁2 )
𝑑𝑧
Here, we are left with an issue as we do not know if NA changes with position or concentration. However,
using the mass balance, assume steady state, constant diffusivity, no generation, 1 dimensional diffusion
This creates a unique issue as the flux is now dependent on itself and the flux of the other species.
and constant pressure and temperature:
Nevertheless, we have more than enough tools to analyse problems relating to diffusion.

0 = − (0 + 0 +
∂𝑧 )
∂𝑁𝐴,𝑍
+0
Analysis of diffusion
Diffusion of species A through a stagnant species B
∂𝑁𝐴,𝑍
=0
If we have a tube contain liquid A, with some of it evaporating through B, which we assume is stationary ∂𝑧
(which is true when the system is in equilibrium), we can make a few simplifications to our equation for
Thus, NA remains constant with changing z. Therefore we can integrate: Since xA is varying with z, xB will also be changing. The mole fractions of both species is varying with
location, thus diffusive flux of B is non-zero, yet absolute flux of B is zero. B is stationary, but has a diffusive
−𝑐𝑇 𝐷 flux. Its diffusive flux exactly cancels out bulk motion.

∫ ∫ 1 − 𝑥𝐴 𝐴
𝑁𝐴 𝑑𝑧 = 𝑑𝑥
Diffusion of species A through a stagnant species B with moving
interface (Stefan tube)
𝑁𝐴 𝑧 = 𝑐𝑇 𝐷 ln(1 − 𝑥𝐴 ) + 𝑐

We could substitute initial conditions to eliminate c. However, if we bounded the integrals, we would get an
expression for NA:

𝑧2 − 𝑧1 ( 1 − 𝑥𝐴1 )
𝑐𝑇 𝐷 1 − 𝑥𝐴2
𝑁𝐴 = ln

We realise that the logarithm can be rewritten as:

𝑧2 − 𝑧1 ( 𝑥𝐵1 )
𝑐𝑇 𝐷 𝑥𝐵2 𝑐𝑇 𝐷 (𝑥𝐴1 − 𝑥𝐴2 )
𝑁𝐴 = ln =
𝑥𝐵,𝑙𝑚 (𝑧2 − 𝑧1 )

𝑥𝐵2 − 𝑥𝐵1 (𝑥𝐴1 − 𝑥𝐴2 )

𝑥𝐵,𝑙𝑚 = =
ln( 𝑥𝐵2
𝐵1 )
ln( 𝑥𝐵2
𝐵1 )
𝑥 𝑥
3. [ENTER DESCRIPTION]

For a short period of time, the interface of A and B will be at the same spot, but as A diffuses into B, the
The above is a simplification is not commonly used. If we integrate between z and z1, we would get:
level of the interface will drop. But, because it is so slow, we can still treat the flux as constant. However, to
analyse this, let us look at the material balance:

𝑧 − 𝑧1 ( 1 − 𝑥𝐴1 )
𝑐𝑇 𝐷 1 − 𝑥𝐴
𝑁𝐴 = ln
𝑑
𝑑𝑡 ( 𝐴,0 )
𝑐 𝑉 = −𝑁𝐴 𝐴
As the flux is the same in the two equations, we will get:

The change in number of moles in the solution will be equivalent to the number diffusing through B.

𝑧 − 𝑧1 ( 1 − 𝑥𝐴1 ) 𝑧2 − 𝑧1 ( 1 − 𝑥𝐴1 )
𝑐𝑇 𝐷 1 − 𝑥𝐴 𝑐𝑇 𝐷 1 − 𝑥𝐴2
ln = ln
ln(
1 − 𝑥𝐴1 )
𝑑 𝑐𝑇 𝐷 1 − 𝑥𝐴2
𝑐𝐴,0 (𝐴(𝐿 − 𝑧)) = −𝐴
𝑑𝑡 𝑧
Using 𝑧1 = 0 and 𝑧2 = 𝐿 :

ln(
1 − 𝑥𝐴1 ) 𝐿 ( 1 − 𝑥𝐴1 )
1 − 𝑥𝐴 𝑧 1 − 𝑥𝐴2 Here, L is the total length of the tube and z is the distance from the top of the tube to the interface (z2-z1).
= ln Note the flux is dependent on the interface itself, so always work with general symbols first. Rearrange and
integrate:

𝑐𝐴,0 𝑧 ( 1 − 𝑥𝐴1 )
𝑐𝑇 𝐷 1 − 𝑥𝐴2
( 1 − 𝑥𝐴1 ) = ( 1 − 𝑥𝐴1 )
1 − 𝑥𝐴 1 − 𝑥𝐴2 𝑧/𝐿 𝑑(𝐿 − 𝑧)
=− ln
𝑑𝑡
 𝐷 (𝑐𝐴2 − 𝑐𝐴1 ) 𝐷 (𝑐𝐴 − 𝑐𝐴1 )
𝑁𝐴 = − =−
𝑧2 − 𝑧1 𝑧 − 𝑧1
∫ 𝑐𝐴,0 𝑧 ( 1 − 𝑥𝐴1 )
𝑐𝑇 𝐷 1 − 𝑥𝐴2
∫
𝑧𝑑𝑧 = ln 𝑑𝑡
 Dividing both sides by 𝑐𝑇 , we obtain a linear profile for the mole fraction 𝑥𝐴 .
𝑋

𝑥𝐴,2 − 𝑥𝐴,1
𝑥𝐴 = 𝑥𝐴,1 + (𝑧 − 𝑧1 )
𝑧2 − 𝑧20 𝑧2 − 𝑧1
= 𝑋𝑡
2
Diffusion and surface reaction
If there was an irreversible very fast catalytic chemical reaction on the surface where all the molecules of A
𝑧2 − 𝑧20 = 2𝑋𝑡
arriving on the surface are immediately converted to the product B:

The level of the interface will slowly get lower as time goes by as expected, but it will do so proportionally
2𝐴 → 𝐵
to 𝑡1/2 .

Equimolar counter diffusion We could visualise the problem as A and B in counter diffusion with A moving towards the catalytic surface
and B moving away. We can assume that the reaction happens spontaneously on the surface:
In this case, equimolar counter diffusion simply means for every mole of A passing in one direction, there
will be one mole of B passing in the opposite direction:

𝑁𝐴 = −𝑁𝐵

This offers a great simplification to Fick’s law:

𝑑𝑥𝐴 𝑑𝑥𝐴
𝑁𝐴 = 𝑥𝐴 (𝑁𝐴 + 𝑁𝐵 ) − 𝑐𝑇 𝐷 = −𝑐𝑇 𝐷 4. [ENTER DESCRIPTION]
𝑑𝑧 𝑑𝑧
An initial simplification we can make:
NA is still a constant from the mass balance, thus:

𝑁𝐴 = −2𝑁𝐵
𝑑𝑥𝐴
𝑁𝐴 = −𝑐𝑇 𝐷
𝑑𝑧
We can also see from the mass balance that, at steady state and the fact that there is no reaction
happening between the catalyst and z2:

∫ ∫
𝑁𝐴 𝑑𝑧 = −𝑐𝑇 𝐷𝑑𝑥𝐴 = − 𝐷𝑑𝑐𝐴
𝑑𝑁𝐴,𝐵
=0
𝑑𝑧

𝐷 (𝑐𝐴2 − 𝑐𝐴1 )
𝑁𝐴 = − Thus NA and NB are both constants throughout, and since NA=-2NB at the surface, we can infer
𝑧2 − 𝑧1 that NA=-2NB everywhere:

This means that the concentration profile is linear and independent of diffusivity for equimolar counter 𝑑𝑥𝐵 𝑑𝑥𝐵
diffusion, and absolute flux is constant. 𝑁𝐵 = 𝑥𝐵 (𝑁𝐴 + 𝑁𝐵 ) − 𝑐𝑇 𝐷 = −𝑥𝐵 𝑁𝐵 − 𝑐𝑇 𝐷
𝑑𝑧 𝑑𝑧
 Using the two expressions together gives a :
𝑑𝑥𝐵
𝑁𝐵 (1 + 𝑥𝐵 ) = −𝑐𝑇 𝐷
𝑑𝑧

𝑑𝑧 ( 𝑑𝑧 )
𝑑 𝑑𝑐𝐴
0=− −𝐷 − 𝑘𝑐𝐴
𝑐𝑇 𝐷
∫ ∫ (1 + 𝑥𝐵 ) 𝐵
𝑁𝐵 𝑑𝑧 = − 𝑑𝑥
𝑑2 𝑐𝐴
𝐷 − 𝑘𝑐𝐴 = 0
𝑑𝑧2

𝐿 ( 1 + 𝑥𝐵2 ) 𝐿 ( 1 + 𝑥𝐵2 )
𝑐𝑇 𝐷 1 + 𝑥𝐵1 𝑐𝑇 𝐷 2
𝑁𝐵 = ln = ln
Solving this ODE using a substitution (note that this is a second order linear ODE with constant coefficients,
which is discussed in Mathematics 1):
This tells us that the rate at which B diffuses is dependent on the diffusivity (only logical) and the length L
(as the shorter distance means A can reach the catalyst faster and produce B sooner). A* NB is also equal
𝑑2 𝑐𝐴
to the rate of reaction as the rate at which it diffuses does depend on the rate at which it is produced. 𝑐𝐴 = 𝑒𝑚𝑧 ; = 𝑚2 𝑒𝑚𝑧
𝑑𝑧2
Diffusion and bulk reaction
𝐷𝑚2 𝑒𝑚𝑧 − 𝑘𝑒𝑚𝑧 = 0
Let us take a new scenario with bacteria located in a bulk where oxygen needs to diffuse through it so that
the bacteria can react with it to decompose the material, assuming steady state, constant diffusivity, 1
dimensional diffusion, constant temperature and pressure and first order reaction:
𝑘 ‾‾𝑘‾
𝑚2 = → 𝑚 = ±√
𝐷 𝐷

Thus, the solution will be:

𝑘 𝑘
𝑧√ 𝐷 −𝑧√ 𝐷
𝑐𝐴 = 𝐶1 𝑒 + 𝐶2 𝑒

The constants can be solved for using the boundary conditions:

5. [ENTER DESCRIPTION]

{ 𝑁𝐴 = 0 at 𝑧 = 𝐿
Essentially, a reaction is happening throughout the diffusion. Since total concentration cT is dominated by
𝑐𝐴 = 𝑐𝐴,0 at 𝑧 = 0
ambient semisolid medium, we may take it as constant. Also. as we are now looking at the diffusion of a
fluid in a solid, where the oxygen is quite dilute, we can assume NA+NB=0 and simplify Fick’s law:
The latter is a no-flux boundary condition due to the impermeable wall.
𝑑𝑥𝐴 𝑑𝑐𝐴
𝑁𝐴 = 𝑥𝐴 (𝑁𝐴 + 𝑁𝐵 ) − 𝑐𝑇 𝐷 = −𝐷
𝑑𝑧 𝑑𝑧 𝑐𝐴,0 = 𝐶1 + 𝐶2 (1)

Use the mass balance:

‾‾𝑘‾ ‾‾𝑘‾
0 = √ 𝐶1 𝑒 √ 𝐷 − √ 𝐶2 𝑒 √ 𝐷 → 0 = 𝐶1 𝑒 √ 𝐷 − 𝐶2 𝑒 √ 𝐷 (2)
𝑘 𝑘 𝑘 𝑘
𝐿 −𝐿 𝐿 −𝐿

𝑑𝑁𝐴 𝐷 𝐷
0=− − 𝑘𝑐𝐴
𝑑𝑧
Multiply Equation (1) by a factor to solve the system of equations:
 𝑘 𝑘 𝑘
𝐿√ 𝐷 𝐿√ 𝐷 𝐿√ 𝐷 Powered by Wiki.js
𝑐𝐴,0 𝑒 = 𝐶1 𝑒 + 𝐶2 𝑒 (1)

𝑘
𝐿√ 𝐷

( )
𝑘 𝑘 𝑘 𝑐𝐴,0 𝑒
𝑒 √𝐷 𝑒 √𝐷 𝑒 √𝐷
𝐿 𝐿 −𝐿
𝑐𝐴,0 = 𝐶2 + → 𝐶2 = 𝑘 𝑘
𝐿√ 𝐷 −𝐿√ 𝐷
𝑒 +𝑒

𝑘
−𝐿√ 𝐷
𝑐𝐴,0 𝑒
∴ 𝑐𝐴,0 − 𝐶2 = 𝐶1 → 𝐶1 = 𝑘 𝑘
𝐿√ 𝐷 −𝐿√ 𝐷
𝑒 +𝑒

𝑘 𝑘
(𝑧−𝐿)√ 𝐷 (𝐿−𝑧)√ 𝐷
𝑐𝐴 𝑒 +𝑒
∴ =
𝑐𝐴,0 𝑘
𝐿√ 𝐷 −𝐿√ 𝐷
𝑘
𝑒 +𝑒

Review Questions
1. Define diffusion and the terminologies involved: local molar concentration, total concentration, molar
average velocity, mole fraction, absolute flux, and diffusive flux.

2. State Fick’s law, explaining what each term represent. Why is there a negative sign?

3. Formulate the mass balance for a control volume over a cuboid of dimensions dx, dy and dz. From this,
derive the diffusion equation in Cartesian coordinates.

4. Rearrange Fick’s law in terms of absolute flux. What is the difference between absolute and diffusive
flux?

5. Derive expressions for the absolute flux of species A diffusing through a static species B, state all the
assumptions involved. Why is the diffusive flux of B non-zero and what does this interpret?

6. For the same set up, consider the level of species A changes with time. Derive an expression for the
level of species A varying with time. Is the steady state assumption reasonable?

7. Derive expressions for the concentration profile and molar flux of equimolar counter diffusion, state all
the assumptions. Does the concentration profile depend on diffusivity?

8. Derive expressions for the concentration profile and molar flux of diffusion with fast surface reaction,
state all the assumptions.

9. Derive expressions for the concentration profile and molar flux of diffusion with first order bulk reaction,
state all the assumptions.

10. Derive expressions for the concentration profile and molar flux of radial diffusion of species A through a
static species B in cylindrical coordinates.