Kinetic processes in materials

Chapter 2. Diffusion

Prof. Nguyen Hong Hai

Hanoi University of Science and Technology
 The study of phase
transformations concerns those
mechanism by which a system
attempts to reach and how long
it takes.
 One of the most fundamental
process that control the rate at
which many transformations
occurs is the diffusion.
 The reason why diffusion occurs
is always so as to produce a
decrease in Gibbs free energy
Fig.2.1. Free energy and chemical
potential changes during diffusion
a) and b) “down-hill” diffusion,
c) and d) “up-hill” diffusion.
 Diffusion ceases when
the composition is
Down-hill everywhere the same
diffusion and the system is in
equilibrium
A2 B1 e) 2A > 1A therefore
B
A atoms move from A
A1 (2) to (1), 1B > 2B
B2
therefore B atoms
B move from (1) to (2). A B
A 2 1 Homogeneous phase

Diffusion ceases when

A1 B2 the compositions of
Up-hill phases are quite
diffusion different

B1 f) 1A > 2A therefore B1

A2 A2
A atoms move from
(1) to (2), 2B > 1B
therefore B atoms
A B A 1 B
2 1 move from (2) to (1). 2
Two separated phases
General rule: Diffusion ceases when the chemical potentials of all atoms are
everywhere the same and the system is in equilibrium.
There are two common mechanisms by which
atoms can diffuse through a solid:
- substitutional atoms usually diffuse by a
vacancy mechanism
- whereas the smaller interstitial atoms
migrate by forcing their way between the
larger atoms. i.e. interstitially.
Vibration energy
Normally a substitutional atom in a crystal
oscillates about a given site and is surrounded
by neighbouring atoms on similar sites.

The mean energy of a vibrational atom, neglecting a constant

term that cannot be changed by temperature, is:
E = KE + PE = 3kBT
The heat capacity per atom will be equal to C = dE/dT = 3kB

Normally the movement of a substitutional atom is limited by

its neighbours and the atom can not move to another site.
However, if an adjacent site is vacant it
can happen that a particularly violent
oscillation result in the atom jumping
over on to the vacancy (fig. 2.2a).

In order for the jump to occur the

shaded atoms (fig. 2.2.b) must move
Fig. 2.2. Movement of an atom into an
apart to create enough space for the adjacent vacancy in an fcc lattice. a) A close-
migrating atom to pass between. packed plan; b) A unit cell showing the four
atoms (shaded) which must move before the
jump can occur.

Therefore the probability that any

atom will be able to jump into a
vacant site depends on the
probability that it can aquire
sufficient vibrational energy.
 The rate at which any given atom is able to migrate
through the solid will be determined by:
1. the frequency with which it encounters a vacancy
2. this in turn depends on the concentration of
vacancies in solid.
 Both the probability of jumping and the
concentration of vacancies are extremely sensitive to
temperature.
 When the solute atom is
appreciably smaller in diameter
than the solvent, it occupies one
of the interstitial site between
the solvent atoms

Fig. 2.3. a) Octahedral interstices

in an fcc crystal; b) Octahedral
interstices in a bcc crystal.
 Usually the concentration of interstitial atoms is so low that only a small fraction of
the available sites is occupied.
 This means that each interstitial atom is always surrounded by vacant sites and can
jump to another as often as its thermal energy permits it to overcome the strain
energy barrier to migration (fig. 2.4).

Fig. 2.4. A 100 plane in an fcc lattice

showing the path of an interstitial
atom diffusing by the interstitial
mechanism.
 The point defect in the crystal lattice
a) Vacancy b) Interstitial atom c) Interstitial impurity
 Consider a simple model of a
dilute interstitial solid solution
(alloy) where the parent atoms
are arranged on a simple cubic
lattice and the solute B atoms fit
perfectly into the interstices
without causing any distortion of
the parent lattice.

 Assume that the solution is so

dilute that every interstitial atom
is surrounded by six vacant
interstitial sites. X

 If the concentration of B varies in one dimension (x) through the solution, the B atoms
can diffuse throughout the material until their concentration is the same everywhere.
 The problem to be considered concerns how this diffusion is related to the random
jump characteristics of the interstitial atoms.
 Consider the exchange of atoms
between two adjacent atomic planes
as (1) and (2).
 Assume that on average an
interstitial atom jumps B times per
second and that each jump is in a
random direction, i.e. there is an
equal probability of the atom
jumping to every one of the six
adjacent sites.
 If plane (1) contains n1 B-atoms per
m2, the number of atoms that will
jump from plane (1) to (2) in 1s (JB)  1 atoms m-2 s-1
will be given by: J B  B n1 (2.1)
6
 During the same time the number of atoms
that jump from (2) to (1), assuming B is
independent of concentration, is given by:
 1
JB  B n 2 atoms m-2 s-1
6
Since n1 > n2 there will be a net flux of atoms The different
from left to right given by: in the
concentration
  1 of B in planes
J B  J B  J B  B n1  n2  (2.2)
(1) and (2)
6
If the separation of planes (1) and (2) is , the
concentration of B at the position of plane (1)
CB(1) = n1/ atoms m-3 and CB(2) = n2/.
Therefore (n1 – n2) = [CB(1) – CB(2)] and it can CB is known as
be seen that x concentration gradient
CB(1) – CB(2) = -(CB /x)
1  C
J B   B 2  B atoms m-2 s-1
Substituting these equations into Equation 2.2. gives: 6  x
 Thus in the presence of a concentration 1  C
gradient the random jumping of individual J B   B 2  B
6  x
atoms produces a net flow of atoms down
the concentration gradient.

Substituting

1  diffusion coefficient
DB   B 2  of B [m2 s-1] (2.3 )
6 

C B
J B   DB [quantity m-2s-1] (2.4)
x

This equation is known as Fick’s first law of diffusion.

 Fick’s first law of diffusion

C B
J B   DB
x
Note

 When the jumping of B atoms is truly random with a frequency independent of

concentration, DB is also a constant independent of concentration.

 If the atomic jumps occur not completely randomly, DB is made to vary with
composition.

 For example the diffusion coefficient D for C in fcc-Fe at 10000 C is 2.5  10-11 m2s-1
at 0.15 wt% C, but 7.7  10-11 m2s-1 at 1.4 wt% C.
 Example: Estimate the jump frequency of a carbon atom in -Fe at 10000C,
given: the lattice parameter of -Fe is ~0.37 nm, D = 2.5 x 10-11 m2s-1.

1 
DB   B 2 
6 
a  0.37 nm
 6D 
B   2B 
  

 = 0.37 / 2 = 0.26 nm   = 2 x 109 jumps s-1

Note: If the vibration frequency of the carbon atoms is ~ 1013, then only about
one attempt in 104 results in a jump from one site to another.
 In the case of random walk of a single diffusing atom (the direction of each new jump
is independent of the direction of the previous jump) in three dimensions it can be
shown that after n steps of length , the ‘average” atom will be displaced by a net
distance  n from its original position.

 Therefore after a time t the average atom will have

r = (t)
advanced a radial distance r from the origin, where
r =   t (2.5)

1 2  6 DB 
Substituting DB   B  B   2 
6    
r = 2.4  Dt (2.6)
r = 2.4 (Dt)

 The distance  Dt is a very important quantity in

diffusion problems.
 In order to move an interstitial atom to an
adjacent interstice the atoms of the parent
lattice must be forced apart into higher
energy positions.
 The work that must be done to accomplish this
process causes an increase in the free energy of
the system by Gm (or Ga ) - Fig. c - and is
known as the activation energy for the migration
of the interstitial atom.
 On average, the fraction of atoms with an
energy of G or more than the mean energy is
given by exp(-G/ RT).

Thus if the interstitial atom in Fig. a is vibrating with a mean frequency v in the x
direction it makes v attempts per second to jump into the next site and the fraction
of these attempts that are successful is given by exp (-G/ RT).
 exp(-G/ RT)

 Now the atom is randomly vibrating in three-dimensional

space, and if it is surrounded by z sites to which it can
jump, the jump frequency is given by:
 Gm
B  zv exp
RT
Combining with 1  1 S   H m (2.8)
D B   B 2
 DB    2 zv exp m  exp
6  6 R  RT
 H a 
Simplified (2.8) to an Arrhenius-type equation: rate exp  
 kT 
Comparing (2.8) with (2.9) gives:

1 S  and  QID
DBO    2 zv exp m  DB  DBO exp (2.9)
6 R  QID = Hm RT
 DB0 is known as the maximal diffusion coefficient (at infinite temperature, m2/s)
 In the case of interstitial diffusion the activation enthalpy for diffusion, QID , is
only dependent on the activation energy barrier to the movement of interstitial
atoms from one site to another.
 Steady-state
conditions

 In most practical situations steady-state

conditions are not established, and Fick’s
law can no longer be used (fig. a)
 The flux at any point along the x-axis will
Steady-state
depend on the local value of DB and CB / x conditions
(Fig.b).
J1 = J2
 In order to calculate how the concentration of
B at any point varies with time consider a J1 > J2
narrow slice of material with an area A and a
thickness x (Fig.c).
 The number of interstitial B atoms that diffuse
into the slice across plane (1) in a small time
interval t will be J1At, and the number of A
atoms that leave is J2At.
 Since J1 > J2 the concentration of B within the
slice will have increased by: J 1  J 2 At
C B 
Ax
 CB 
J1  J 2 At
Ax

In the limit as C B J B
t  0 
t x
Substituting Fick’s first law

C B
J B   DB
x

C B   C B 
  DB 
t x  x  Fick’s second law
 Fick’s second law
 If variation of DB with concentration can
C B   C B 
be ignored this equation can be simplified   DB 
t x  x 

 This equations relate the rate of C B  2C B

change of composition with time to  DB (2.16)
the concentration profile CB(x).
t x 2
 2CB/x2 is the curvature of CB versus
x distance curve.
 If the concentration profile appears as
shown in Fig. a it has a positive
curvature everywhere and the
concentration at all points on such
curve will increase with time (CB/ t
positive) and vice versa (Fig. b).
 Two solutions to the diffusion
equation are of practical importance:
homogenization heat treatment

and carburization of steel.

For example for the case of carbon diffusion in

austenite at 10000 C, D  4 x 10-11 m2s-1, which
means that a carburized layer 0.2 mm thick
requires a time of (0.2 x 10-3)2/ 2.42  4  10-11,
i.e. ~ 700s.
r = 2.4  Dt 0.2 mm
 In substitution diffusion an atom can only

jump if there happens to be a vacant site at

one of the adjacent lattice positions.

 The simplest case of substitutional diffusion is the self-diffusion of

atoms in a pure metal.

 The rate of self-diffusion can be measured
experimentally by introducing a few radioactive A
atoms (A*) into pure A and measuring the rate at
which penetration occurs at various temperatures.
 Since A* and A atoms are chemically identical their
jump frequencies are also almost identical.

 Thus the diffusion coefficient can be related to

the jump frequency by Equation 2.3, that is:

1 2
D A*  D A    (2.22)
6

 In the case of substitutional diffusion the difference is

that once an atom has jumped into a vacancy the next
jump is not equally probable in all directions, but is
most likely to occur back into the same vacancy.
 The probability that any attempt at jumping
will be successful is given by exp(-Gm/RT)
as in the case of interstitial migration.
 However, most of time the adjacent site will
not be vacant and the jump will not be possible.
 The probability that an adjacent site is vacant is
given by zXV, where z is the number of nearest
neighbours and XV is the probability that any
one site is vacant, which is just the mole
fraction of vacancies in the metal.
 Combining all these probabilities gives the probability
of a successful jump as zXV exp (-Gm/ RT).
 Since the atoms are vibrating with a temperature-independent frequency v,
the number of successful jumps any given atom will make in 1s is given by:

 Gm
  vzX v exp (2.23)
RT
 If the vacancies are in thermodynamic Combine with:
equilibrium:
Xv 

 Gv  Gm 1
X v  X ve  exp   vzX v exp D A*  D A   2 
RT RT 6
(2.23) (2.22)

1  Gm  Gv 
DA   2vz exp
6 RT Substituting G = H - TS

1 2
DA   vz exp
S m  S v   H m  H v 
exp   (2.26)
6 R  RT 
 (2.26)

 For most metals vibration frequency, v, is 1013.

 In fcc metals, for example, the number of nearest neighbours, z = 12
and the jump distance  = a2
 so (2.26) can be written more concisely as

 QSD 1 2
D0   vz exp
Sm  Sv 
D A  D0 exp (2.27) with
RT 6 R
QSD = Hm + Hv

QSD: activation enthalpy for self diffusion

 Comparison substitutional diffusion with interstitial

 QSD  Q ID (2.9)
D A  D0 exp (2.27) D B  D BO exp
RT RT

1 2 Sm  Sv  1 2
   zv exp
S m 
D0   zvexp D BO
6 R 6 R 
QSD = Hm + Hv QID = Hm

Equation 2.27 is the same as was obtained for interstitial diffusion except that the
activation energy for self-diffusion has an extra term (HV), because self-diffusion
requires the presence of vacancies whose concentration depends on HV.
 For most close-parked metals (fcc and hcp) the activation enthalpy for self diffusion,
QSD, is proportional to the equilibrium melting temperature and QSD/RTm ~ 18;
 This means increasing the interatomic bond strength (i.e. QSD  ) makes the process of
melting more difficult; that is, Tm is raised
 The interatomic bond strength also makes diffusion more difficult.
 The diffusivity at the melting temperature D(Tm) ~ 1 m2 s-1.
 Consequence: the diffusion coefficient of all materials with a given crystal structure
and bond type will be approximately the same at the same fraction of their melting
temperature, i.e.
D (T/Tm) = constant
Class Kinetic
Metalprocesses
Tm, Kin Materials
Q, kJ/mol Q/RT D0, mm2/s D0(Tm ) mm2/s

bcc -Ce 1071 90 10.1 1.2 49

Rare
Earth -Yb 1796 121 8.1 1.2 3600

bcc Na 371 43.8 14.2 24.2 16

Alkali
metals Li 454 55.3 14.7 23 9.9

bcc -Fe 1811 239.7 15.9 200 26

Transitio
n metals -Ti 1933 251.2 15.6 109 18

hcp Zn 692 91.6 15.9 II c 13 1.6

Mg 922 136 17.8  c 150 2.9

Al 933 142.0 18.3 170 1.9

fcc
Cu 1356 200.3 17.8 31 0.59

Ni 1726 279.7 19.5 190 0.65

-Co 1768 283.4 19.3 83 0.35

-Fe 1805 284.1 18.9 49 0.29

tet -Sn 505 107.1 25.5 II c 770 0.0064

Diamon Ge 1211 324.5 32.3 440 4.4 x 10-5

d cubic
Si 1683 496.0 35.5 0.9 x 106 3.6 x 10-4
Example : effect of temperature on self-diffusion in Cu.

 At 8000 C (1073 K) DCu = 5 x 10-9 mm2s-1. The jump distance  in Cu is 0.25 nm and
Equation 2.3 gives Cu = 5 x 105 jumps s-1.

1  r = 2.4 (Dt)
DB   B 2  (2.3)
6 

After an hour at this temperature (Dt) ~ 4 m.

 At 200C DCu ~ 10-34 mm2s-1, i.e.  ~ 10-20 jumps s-1.

Alternatively, each atom would make one jump every 1012 years!
 The jumping of atoms into vacant sites can equally
well be as the jumping of vacancies onto atom sites.

 However, a vacancy is always surrounded by sites to

which it can jump and it is thus analogous to an
interstitial atom.

1 
DB   B 2 
6 

Therefore a vacancy can be considered to have 1

its own diffusion coefficient given by : Dv  v 2

6
 1 2 Sm   Hm
By analogy with the case of
DB    zvexp  exp (2.8)
6 R  RT
intertitial diffusion
1 2 Sm   Hm
Comparing 2.31 and 2.26 Dv    zvexp  exp (2.31)
6 R  RT

1 2 S m  S v   H m  H v 
D A   zv exp exp   (2.26)
6 R  RT 

Xve : mole fraction of

Dv  D A / X e
v vacancies

Dv is many orders of magnitude greater than DA , the diffusivity of substitutional atoms.

During self-diffusion all atoms are chemically indentical, thus the probability of
finding a vacancy adjacent to any atom and the probability that atom will make a
jump into the vacancy is equal for all atoms.
This leads to a simple relationship between jump frequency and diffusivity:
A-atom B-atom
1
D A*  D A   2 
6

In binary substitutional alloys, the rate at which solvent (A)

solute (B) atoms can move into a vacant site is not equal.

Each atomic species must be given its own intrinsic diffusion

coefficient, DA và DB, and they are defined such that Fick’s
first law applies to diffusion relative the lattice, that is:
Where JA và JB are the fluxes
C A C B of A and B across a given
J A   DA J B   DB (2.32)
lattice plane.
x x
In order to derive Fick’s secon law let’s
consider the interdiffusion of A and B in a
diffusion couple that is made by welding
together blocks A và B.

If the couple is annealed at high enough

temperature, a concentration profile will
develop as shown.

To simplify assume that the total number of

atoms per unit volume is a constant, C0 ,
independent of composition. Then:

C0 = CA + CB

and
C A C B
J A   DA J B   DB
x x Assume DA > DB so JA  JB
When atoms migrate by the vacancy process
the jumping of an atom into a vacant site can
equally well be regarded as the jumping of
the vacancy onto atom in the opposite
direction (- JA và – JB ).

As JA  JB

there will be a net flux of vacancies:

JV = - JA - JB (2.33)

C A
J V  DA  DB  (2.34)
x
In order to maintain the vacancy
concentration everywhere near
equilibrium, vacancies must be
created on the B-rich side and
destroyed on the A-rich side
(figure c).

The rate at which vacancies are

created or destroyed at any point
is given by Eq. 2.35

 CV J V
  (2.35)
t x
 Shringking Growing
plane plane

Absorbed
It is the net flux of vacancies vacancy
across the middle of the
diffusion couple that give rise
to movement of the lattice.

Vacancies can be absorbed by

the extra half-plane of the
edge dislocation shrinking
The formation of vacancy:
while growth of the plane (a) atom from site 1 moves to interstice (2),
so dislocation moves to above slide
can occur by the emmission
plane, and at position (2) is ocuppied by
of vacancies. a interstitial atom
(b) Atom moves from position (1) to position
(2), left a vacancy at position (2)
If this mechanism operates on each side
of the diffusion couple then the required
flux of vacancies can be generated.

This means that extra atomic planes

will be introduced on the B-rich side A B
while whole planes of atoms will be
“eaten” away on the A-rich side.

Consequently the lattice planes in the

middle of the couple will be shifted to
the left.
The velocity at which any given lattice plane
moves, v, can be related to the flux of
vacancies crossing it.
If the plane has an area A, during a small
time interval t the plane will sweep out a
volume of Av.t containing Av.t.C0 atoms.

This number of atoms is removed by the

total number of vacanvies crossing the plane
in the same time interval, JV A. t, giving:

Av.t.C0 = JV A. t JV = C0  v (2.36)

Thus the velocity of the lattice planes, v, will vary across the couple in the same way as JV.

C A
Substituting JV from JV  DA  DB  (2.34) into 2.36 gives:
x
X A
v  DA  DB  (2.37) where XA is mole fraction of A,
x XA = C A / C 0
 In practice, more important questions
concern how long homogenization of
an alloy takes, or how rapidly the
composition will change at a fixed
Consider a thin slice of materials x at fixed position  Fick’s 2
distance x from one end of the couple of
which is outside the diffusion zone.
If the total flux of A atoms entering this slice
across plane 1 is J’A and the total flux leaving
is J’A + (J’A / x)x the same arguments as J’A J A'
J 
'
A x
were used to derive Equation x

CV JV
  (2.35)
t x
can be used C A J 'A 1 2
  (2.38)
to show that: t x x x
 The total flux of A across a stationary plane is the sum of two contributions:
C A
i) A diffusive flux: J A   DA due to diffusion relative to the lattice
x
ii) A flux vCA due to the velocity of the lattice in which diffusion is occuring
C A
J 'A   DA  vC A (2.39)
x
 C A X A
Combining J ' A   DA  vC A (2.39) with v  DA  DB  (2.37)
x x
C A
J ' A   X B DA  X A DB  (2.40)
x
where XA = CA / C0 và XB = CB / C0 are mole fraction of A and B.

This can be simplified by defining interdiffusion

coefficient, Di as:

Di = (XB DA + XADB ) (2.41)

C A
so that Fick’s first law become: J ' A   Di (2.42)
x
CB C A
Likewise J ' B   Di  Di i.e. J’B = - J’A
x x
 J ' A   Dki
C A C A J 'A
Substitute (2.43) into   (2.38)
x t x
C A   C A  (2.43)
  Di 
t x  x 
This equation is Fick’s second law for diffusion in substitutional alloy.

The only difference with the interstitial

diffusion is that the interdiffusion C B  2C B
coefficient Di depends on DA and DB,
 DB
t x 2
whereas in interstitial diffusion DB alone
is needed.

X A
The equations 
v  DA  DB  (2.37) and Di = (XB DA + XADB ) (2. 41)
x are known as Darken’s Equations
The variation of Di with composition can be estimated in cases where it has not
been measured, by utilizing two experimental obsevations:
1. For a given crystal structure, Di at the melting point is roughly constant.
Therefore if adding B to A decreases the melting point, Di will increase at the
given temperature, and vice versa.
2. For a given solvent and temperature, both interstitial and substitutional
diffusion are more rapid in a bcc lattice than a close-packed lattice.
For example, for the diffusion of carbon in Fe at 910 0C, DC / DC  100.
At 850 0C the self-diffusion coefficient for Fe are such that: DFe / DFe  100.

The reason for this difference lies in

the fact that the bcc structure is more
open and the diffusion processes
require less lattice distortion.
In dilute substitutional alloys XA  1 và XB  0, thus:
Di = DB
This means that the rate of homogenization in dilute alloys is controlled by how fast
the solute B atoms can diffuse.
It is often found that DB in a dilute solution of B in A is greater than DA .
The reason for this is that the solute atoms can attract vacancies so that there is more
than a random probability of finding a vacancy next to a solute atom with the result
that they can diffuse faster the solvent.
An attraction between a solute atom and a
vacancy can arise if the solute atom is larger
then the solvent atoms or if it has higher
valency.
If the binding energy is very large the vacancy
will be unable to “escape” from the solute
atom. In this case the solute-vacancy pair can
diffuse through the lattice together.
 Element Atomic Number Valency
Valency of Hydrogen 1 1
Valency of Helium 2 0
Valency of Lithium 3 1
Valency of Beryllium 4 2
Valency of Boron 5 3
Valency of Carbon 6 4
Valency of Nitrogen 7 3
Valency of Oxygen 8 2
Valency of Fluorine 9 1
Valency of Neon 10 0
Valency of Sodium (Na) 11 1
Valency of Magnesium (Mg) 12 2
Valency of Aluminium 13 3
Valency of Silicon 14 4
Valency of Phosphorus 15 3
Valency of Sulphur 16 2
Valency of Chlorine 17 1
Valency of Argon 18 0
Valency of Potassium (K) 19 1
Valency of Calcium 20 2
Valency of Scandium 21 3
Valency of Titanium 22 4
Valency of Vanadium 23 5,4
Valency of Chromium 24 2
Valency of Manganese 25 7, 4, 2
Valency of Iron (Fe) 26 2, 3
Valency of Nickel 27 3, 2
Valency of Cobalt 28 2
Valency of Copper (Cu) 29 2, 1
Valency of Zinc 30 2
Fick's first law is based on the assumption that diffusion
C B
eventually stops, that is equilibrium is reached, when the J B   DB
concentration is the same everywhere.
x
This situation is never true in practice because real materials always contain lattice
defects such as grain boundaries, phase boundaries and dislocations. Some atoms can
lower their free energies if they migrate to such defects and at 'equilibrium' their
concentrations will be higher in the vicinity of the defect than in the matrix.
Diffusion in the vicinity of these defects is therefore affected by both the
concentration gradient and the gradient of the interaction energy. Fick's law alone
is insufficient to describe how the concentration will vary with distance and time.
As an example consider the case of a solute atom
that is too big or too small in comparison to the
space available in the solvent lattice. The potential
energy of the atom will then be relatively high due
to the strain in the surrounding matrix.

However, this strain energy can be reduced if the atom is located in a position where it
better matches the space available, e.g. near dislocations and in boundaries, where the
matrix is already distorted.
Segregation of atoms to grain boundaries, interfaces and dislocations is of great
technological importance. For example the diffusion of carbon or nitrogen to
dislocations in mild steel is responsible for strain ageing and blue brittleness.
The segregation of impurities such as Sb, Sn, P and As to grain boundaries in low-alloy
steels produces temper embrittlement.
Segregation to grain boundaries affects the mobility of the boundary and has
pronounced effects on recrystallization, texture and grain growth. Similarly the rate at
which phase transformations occur is sensitive to segregation at dislocations and
interfaces.

The problem of atom migration can be solved by considering the thermodynamic

condition for equilibrium; namely that the chemical potential of an atom must be the
same everywhere: A = A = A …, B = B = B …
Diffusion continues in fact until this condition is satisfied. Therefore it seems reasonable
to suppose that in general the flux of atoms at any point in the lattice is proportional to
the chemical potential gradient. Fick's first law is merely a special case of this more
general approach.
An alternative way to describe a flux of atoms is to consider a net drift velocity (v)
superimposed on the random jumping motion of each diffusing atom.
The drift velocity is simply related to the diffusive flux via the equation:

JB = vB CB (2.44)

Since atoms always migrate so as to remove differences in chemical potential, it is

reasonable to suppose that the drift velocity is proportional to the local chemical
potential gradient:

(2.45)

where MB is a constant of proportionality known as the atomic mobility.

Since B has units of energy, the derivative of B with respect to distance (dB /dx) is
effectively the chemical 'force' causing the atom to migrate.
Combining Equations 2.44 and 2.45 gives:
 (2.46)

 For ideal or dilute solutions (XB  0) B is a constant:
DB = MBRT (2.47)
B : the activity coefficient of B
 For non-ideal concentrated solutions: aB : the activity of B

Thermodynamic factor is
the same for both A and
B and is simply related to
the curvature of the
molar free energy-
thermodynamic factor composition curve.

 When diffusion occurs in the presence of a strain energy gradient, for example,
the expression for the chemical potential can be modified to include the effect
of an elastic strain energy term E, which depends on the position (x) relative to
a dislocation:
B = GB + RT In BXB + E
 In Section 2.5.1 the diffusion of atoms towards or away from dislocations, interfaces,
grain boundaries and free surfaces was considered.
 In this section diffusion along these defects will be discussed.
 All of these defects are associated with a more open structure and it has been shown
experimentally that the jump frequency for atoms migrating along these defects is
higher than that for diffusion in the lattice.

 It will become apparent that under certain circumstances diffusion along these
defects can be the dominant diffusion path.
It is found experimentally that diffusion along grain boundaries and free surfaces
can be described by:
where Db and Ds : the grain boundary and surface diffusivities
DbO and Ds0 are the frequency factors
Qb and Qs are the experimentally determined values of the
activation energies for diffusion.

In general, at any temperature the magnitudes of Db and Ds relative to the

diffusivity through defect-free lattice Dl are such that:
Ds > Db > Dl

This mainly reflects the relative ease with which atoms can migrate along free
surfaces, interior boundaries and through the lattice.
Surface diffusion can play an important role in many metallurgical phenomena, but in
an average metallic specimen the total grain boundary area is much greater than the
surface area so that grain boundary diffusion is usually most important.
 The effect of grain boundary
diffusion can be illustrated by
considering a diffusion couple
made by welding together
two metals, A and B, as shown
in Fig. beside.
o A atoms diffusing along the
boundary will be able to
penetrate much deeper than
atoms which only diffuse
through the lattice.
o In addition, as the
concentration of solute builds
up in the boundaries atoms will
also diffuse from the boundary
into the lattice.
 The process can be compared to
the conduction of heat through a
plastic in which a continuous
network of aluminum sheets is
embedded.
 The temperature at any point in
such a specimen would be
analogous to the concentration
of solute in the diffusion couple.
 Points in the lattice close to grain
boundaries can receive solute via
the “high conductivity path”
much more rapidly than if the
boundaries were absent.

 Rapid diffusion along the grain boundaries increases the mean concentration in a slice
such as dx in Fig. above and thereby produces an increase in the apparent diffusivity in
the material as a whole.
Consider now under what conditions grain boundary diffusion is important.

For simplicity let us take a case of steady-state

diffusion through a sheet of material in which the
grain boundaries are perpendicular to the sheet.

Assuming that the concentration gradients in the

lattice and along the boundary are identical, the
fluxes of solute through the lattice Jl and along the
boundary Jb will be given by:

Db > Dl

However the contribution of grain boundary

diffusion to the total flux through the sheet will
Combined lattice and boundary
depend on the relative cross-sectional areas fluxes during steady-state diffusion
through which the solute is conducted. through a thin slab of material
If the grain boundary has an effective thickness 
and the grain size is d, the total flux will be given by:

Thus the apparent diffusion coefficient in this case:

Dapp = Dl + Db / d

𝒂𝒑𝒑 𝒃
or
𝒍 𝒍
It can be seen that the relative importance of
lattice and grain boundary diffusion depends
on the ratio Db/ Dld.
When Db >> Dld diffusion through the lattice
can be ignored in comparison to grain
boundary diffusion.
 Thus grain boundary diffusion makes a significant
contribution to the total flux when Db > Dld
 Db > Dld
The effective width of a grain boundary is ~
0.5 nm. Grain sizes on the other hand can vary
from ~ 1 to 1000 m (2000  2000000 times
greater) and the effectiveness of the grain
boundaries will vary accordingly.
Greater
The relative magnitudes of Db and Dld are difference at
most sensitive to temperature. lower
temperature
Note that although Db > Dl at all temperatures
the difference increases as temperature
decreases. This is because the activation energy
for diffusion along grain boundaries (Qb) is
Diffusion in a polycrystalline metal
lower than that for lattice diffusion (Q!).

For example, in fcc metals it is generally found that Qb ~ 0.5 Ql .

This means that when the grain boundary diffusivity is scaled by the factor /d
(Dapp = Dl + Db / d ) the grain boundary contribution to the total, or apparent,
diffusion coefficient is negligible in comparison to the lattice diffusivity at high
temperatures, but dominates at low temperatures.
 Db > Dld

In general it is found that grain boundary

diffusion becomes important below about
0.75-0.8 Tm, where Tm is the equilibrium
melting temperature in degrees Kelvin.
Note that the rate at which atoms diffuse
along different boundaries is not the same,
but depends on the atomic structure of the
individual boundary.
This in turn depends on the orientation of
the adjoining crystals and the plane of the
boundary.
Diffusion in a polycrystalline metal
Also the diffusion coefficient can vary with
direction within a given boundary plane.
If grain boundary diffusion is compared to the conduction of heat through a material
made of sheets of aluminium in a plastic matrix, the analogy for diffusion along
dislocations would be aluminium wires in a plastic matrix.
The dislocations effectively act as pipes along which atoms can diffuse with a diffusion
coefficient Dp . The contribution of dislocations to the total diffusive flux through a
metal will of course depend on the relative cross-sectional areas of pipe and matrix.

Dislocations act as a high

conductivity path through
the lattice.

Using the simple model illustrated

in Fig. above it can easily be shown
that the apparent diffusivity
where g is the cross-
through a single crystal containing
𝒂𝒑𝒑 𝒑 sectional area of 'pipe'
dislocations, Dapp , is related to the
𝒍 𝒍 per unit area of matrix.
lattice diffusion coefficient by:
 𝒂𝒑𝒑 𝒑

𝒍 𝒍

In a well-annealed material there are roughly 105 dislocations mm-2.

Assuming that the cross-section of a single pipe accommodates about 10 atoms while
the matrix contains about 1013 atoms mm-2, makes g = 10-7.

At high temperatures diffusion through the lattice is rapid and gDp/ D1 is very small so
that the dislocation contribution to the total flux of atoms is negligible.
However, since the activation energy for pipe diffusion is less than for lattice diffusion,
D1 decreases much more rapidly than Dp with decreasing temperature, and at low
temperatures gDp / DI can become so large that the apparent diffusivity is entirely due
to diffusion along dislocations.