CHAPTER 5

DIFFUSION
PROBLEM SOLUTIONS
5.1 Self-diffusion is atomic migration in pure metals--i.e., when all atoms exchanging positions are of the
same type. Interdiffusion is diffusion of atoms of one metal into another metal.
5.2 Self-diffusion may be monitored by using radioactive isotopes of the metal being studied. The motion
of these isotopic atoms may be monitored by measurement of radioactivity level.
5.3 (a) With vacancy diffusion, atomic motion is from one lattice site to an adjacent vacancy. Selfdiffusion and the diffusion of substitutional impurities proceed via this mechanism. On the other
hand, atomic motion is from interstitial site to adjacent interstitial site for the interstitial diffusion
mechanism.
(b) Interstitial diffusion is normally more rapid than vacancy diffusion because: (1) interstitial atoms,
being smaller, are more mobile; and (2) the probability of an empty adjacent interstitial site is greater
than for a vacancy adjacent to a host (or substitutional impurity) atom.
5.4 Steady-state diffusion is the situation wherein the rate of diffusion into a given system is just equal to
the rate of diffusion out, such that there is no net accumulation or depletion of diffusing species--i.e.,
the diffusion flux is independent of time.
5.5 (a) The driving force is that which compels a reaction to occur.
(b) The driving force for steady-state diffusion is the concentration gradient.
5.6 This problem calls for the mass of hydrogen, per hour, that diffuses through a Pd sheet. It first
becomes necessary to employ both Equations (5.1a) and (5.3). Combining these expressions and
solving for the mass yields

M = JAt =

85

= 2.6 x 10

-3

kg/h

3
5.7 We are asked to determine the position at which the nitrogen concentration is 2 kg/m . This problem
is solved by using Equation (5.3) in the form

J =

3
If we take C to be the point at which the concentration of nitrogen is 4 kg/m , then it becomes
A
necessary to solve for x , as
B

C
xB = x A + D A

Assume x

is zero at the surface, in which case

xB = 0 + 6 x 10 -11 m2 /s

-3
= 1 x 10 m = 1 mm
5.8 This problem calls for computation of the diffusion coefficient for a steady-state diffusion situation.
3
Let us first convert the carbon concentrations from wt% to kg C/m using Equation (4.9a). For 0.012
wt% C
''
CC
=

CC

0.012

CC

0.012

2.25 g / cm3

86

0.944 kg C/m

Similarly, for 0.0075 wt% C

CC'' =

0.0075

0.0075

2.25 g / cm3

= 0.590 kg C/m

Now, using a form of Equation (5.3)

D =

= 3.95 x 10

-11

2
m /s

5.9 This problems asks for us to compute the diffusion flux of hydrogen gas through a 1-mm thick plate of
iron at 250C when the pressures on the two sides are 0.15 and 7.5 MPa. Ultimately we will employ
Equation (5.3) to solve this problem.

However, it first becomes necessary to determine the

concentration of hydrogen at each face using Equation (5.11). At the low pressure (or B) side

CH(B) = 1.34 x 10 -2

-6
9.93 x 10 wt%
Whereas, for the high pressure (or A) side

CH(A) = 1.34 x 10 -2

87

-5
7.02 x 10 wt%
We now convert concentrations in weight percent to mass of hydrogen per unit volume of solid. At
-6
-9
face B there are 9.93 x 10 g (or 9.93 x 10 kg) of hydrogen in 100 g of Fe, which is virtually pure
3
iron. From the density of iron (7.87 g/cm ), the volume iron in 100 g (V ) is just
B

VB =

100 g

7.87 g / cm 3

= 12.7 cm

= 1.27 x 10

-5

Therefore, the concentration of hydrogen at the B face in kilograms of H per cubic meter of alloy [
''
CH(B)
] is just

''
CH(B)
=

9.93 x 10

CH(B)
VB

At the A face the volume of iron in 100 g (V ) will also be 1.27 x 10

-5

''
CH(A)
=

7.02 x 10

3
m , and

CH(A )
VA

Thus, the concentration gradient is just the difference between these concentrations of hydrogen
divided by the thickness of the iron membrane; that is
''
''
CH(B)
CH(A)
C
=
x
xB xA

7.82 x 10

88

At this time it becomes necessary to calculate the value of the diffusion coefficient at 250C using
Equation (5.8). Thus,

D = D o exp

= 1.4 x 10

= 6.41 x 10

-9

2
m /s

And, finally, the diffusion flux is computed using Equation (5.3) by taking the negative product of this
diffusion coefficient and the concentration gradient, as

J =
=

5.10 It can be shown that

Cx =

B
exp

Dt

is a solution to

C
2C
= D
t
x 2
simply by taking appropriate derivatives of the C expression. When this is carried out,
x
2

C
C
B
= D
=
2
1/
t
x
2 D 2t 3 / 2

5.11

x2

2 Dt 1
exp

x2

4Dt

We are asked to compute the diffusion time required for a specific nonsteady-state diffusion
situation. It is first necessary to use Equation (5.5).

89

C x

wherein, C = 0.45, C = 0.20, C = 1.30, and x = 2 mm = 2 x 10

x
o
s
C x

-3

m. Thus,

or
x
erf

2 Dt

By linear interpolation from Table 5.1

z

erf(z)

0.85

0.7707

0.7727

0.90

0.7970

From which

z = 0.854 =

x
2 Dt

Now, from Table 5.2, at 1000C (1273 K)

D = 2.3 x 10 -5 m2 /s

= 1.93 x 10
Thus,

90

-11

2
m /s

2 x 10

0.854 =

Solving for t yields

4
t = 7.1 x 10 s = 19.7 h
5.12 This problem asks that we determine the position at which the carbon concentration is 0.25 wt%
after a 10-h heat treatment at 1325 K when C = 0.55 wt% C. From Equation (5.5)
o
C x

Thus,
x
erf

2 Dt

Using data in Table 5.1 and linear interpolation

z

erf (z)

0.40

0.4284

0.4545

0.45

0.4755

And,
z = 0.4277
Which means that

x
2 Dt

= 0.4277

And, finally

91

x = 2(0.4277)

4.3 x 10

Dt = (0.8554)

= 1.06 x 10

-3

m = 1.06 mm

5.13 This problem asks us to compute the nitrogen concentration (C ) at the 1 mm position after a 10 h
x
diffusion time, when diffusion is nonsteady- state. From Equation (5.5)
C x

= 1

= 1 - erf (0.527)
Using data in Table 5.1 and linear interpolation
z

erf (z)

0.500

0.5205

0.527

0.550

0.5633

0.527

from which
y = erf (0.527) = 0.5436
Thus,

Cx

This expression gives

92

C = 0.046 wt% N
x
5.14 (a) The solution to Fick's second law for a diffusion couple composed of two semi-infinite solids of
the same material is as follows:
C1
C x =

for the boundary conditions
C = C for x < 0, and t = 0
1

C = C for x > 0, and t = 0

2

(b) For this particular silver-gold diffusion couple for which C = 5 wt% Au and C = 2 wt% Au, we
1

are asked to determine the diffusion time at 750C that will give a composition of 2.5 wt% Au at the
50 m position. Thus, the equation in part (a) takes the form

5
2.5 =

It now becomes necessary to compute the diffusion coefficient at 750C (1023 K) given that D = 8.5
o
-5 2
x 10 m /s and Q = 202,100 J/mol. From Equation (5.8) we have
d

D = D oexp

= 8.5 x 10

-5

2
m /s

= 4.03 x 10

-15

2
m /s

Substitution of this value into the above equation leads to

93

5

2.5 =

This expression reduces to the following form:

393.8 s
0.6667 = erf

Using data in Table 5.1 and linear interpolation

z

erf (z)

0.650

0.6420

0.6667

0.700

0.6778

from which
y = 0.6844 =

393.8

And, solving for t gives

5
t = 3.31 x 10 s = 92 h
5.15 This problem calls for an estimate of the time necessary to achieve a carbon concentration of 0.45
wt% at a point 5 mm from the surface. From Equation (5.6b),
x2
= constant
Dt
But since the temperature is constant, so also is D constant, and
x2
= constant
t
or
94

x2

x2
1 = 2
t1
t2

Thus,

(2.5 mm)
10 h

(5.0 mm)
t2

from which
t = 40 h
2
5.16 We are asked to compute the diffusion coefficients of C in both and iron at 900C. Using the
data in Table 5.2,

= 1.69 x 10

-10

2
m /s

= 5.86 x 10

-12

2
m /s

The D for diffusion of C in BCC iron is larger, the reason being that the atomic packing
factor is smaller than for FCC iron (0.68 versus 0.74); this means that there is slightly more
interstitial void space in the BCC Fe, and, therefore, the motion of the interstitial carbon atoms occurs
more easily.
5.17 This problem asks us to compute the magnitude of D for the diffusion of Zn in Cu at 650C (923 K).
From Table 5.2

D = 2.4 x 10-5 m2 /s

95

= 4.8 x 10

-16

2
m /s

5.18 We are asked to calculate the temperature at which the diffusion coefficient for the diffusion of Cu in
-17 2
Ni has a value of 6.5 x 10
m /s. Solving for T from Equation (5.9a)
T =

and using the data from Table 5.2 for the diffusion of Cu in Ni
T =

= 1152 K = 879C
5.19 For this problem we are given D and Q for the diffusion of Cr in Ni, and asked to compute the
o
d
-14 2
temperature at which D = 1.2 x 10
m /s. Solving for T from Equation (5.9a) yields
T =

Qd
R ln D o

272, 000 J/mol

(8.31 J/mol - K) ln 1.1 x 10 -4

= 1427 K = 1154C
5.20 In this problem we are given Q

for the diffusion of Cu in Ag (i.e., 193,000 J/mol) and asked to

compute D at 1200 K given that the value of D at 1000 K is 1.0 x 10

necessary to solve for D from Equation (5.8) as
o

Q
Do = D exp d
RT

96

-14

2
m /s. It first becomes

= 1.0 x 10

-14

2
m /s

= 1.22 x 10

-4

2
m /s

Now, solving for D at 1200 K gives

D = 1.22 x 10

-4

2
m /s

= 4.8 x 10

-13

2
m /s

5.21 (a) Using Equation (5.9a), we set up two simultaneous equations with Q and D as unknowns.
d
o
-16
Solving for Q in terms of temperatures T and T (1273 K and 1473 K) and D and D (9.4 x 10
d
1
2
1
2
-14 2
and 2.4 x 10
m /s), we get

Q d =

= 252,400 J/mol
Now, solving for D from Equation (5.8)
o

Q
d
Do = D1exp
RT

= 9.4 x 10

-16

2
m /s

97

= 2.2 x 10

-5

2
m /s

(b) Using these values of D and Q , D at 1373 K is just

o
d

D = 2.2 x 10

-5

2
m /s

= 5.4 x 10

-15

2
m /s

5.22 (a) Using Equation (5.9a), we set up two simultaneous equations with Q and D as unknowns.
d
o
Solving for Q in terms of temperatures T and T (873 K [600C] and 973 K [700C]) and D and
d
1
2
1
-14
-13 2
D (5.5 x 10
and 3.9 x 10
m /s), we get
2

Q d =

= 138,300 J/mol
Now, solving for D from Equation (5.8)
o

Q
d
Do = D1exp
RT

= 5.5 x 10

-14

2
m /s

= 1.05 x 10

98

-5

2
m /s

(b) Using these values of D and Q , D at 1123 K (850C) is just

o
d

D = 1.05 x 10

-5

2
m /s

= 3.8 x 10

-12

2
m /s

5.23 This problem asks us to determine the values of Q and D for the diffusion of Au in Ag from the
d
o
plot of log D versus 1/T. According to Equation (5.9b) the slope of this plot is equal to - Q /2.3R
d
(rather than - Q /R since we are using log D rather than ln D) and the intercept at 1/T = 0 gives the
d
value of log D . The slope is equal to
o

slope =

-3
-3 -1
Taking 1/T and 1/T as 1.0 x 10 and 0.90 x 10 K , respectively, then the values of log D and
1
2
1
log D are 14.68 and 13.57, respectively. Therefore,
2

Q d =

= 212,200 J/mol
Rather than trying to make a graphical extrapolation to determine D , a more accurate value is
o
obtained analytically using Equation (5.9b) taking a specific value of both D and T (from 1/T) from the
-14 2
-3
plot given in the problem; for example, D = 1.0 x 10
m /s at T = 1064 K (1/T = 0.94 x 10 ).
Therefore

99

Q d
Do = D exp
RT
= 1.0 x 10

-14

2
m /s

= 2.65 x 10

5.24

-4

2
m /s

This problem asks that we compute the temperature at which the diffusion flux is 6.3 x 10
2
kg/m -s. Combining Equations (5.3) and (5.8) yields

J =

-10

Solving for T from this expression leads to

T = d
R

80, 000 J / mol
=

8.31 J / mol - K

= 900 K = 627C
5.25 In order to solve this problem, we must first compute the value of D from the data given at 1200C
o
(1473 K); this requires the combining of both Equations (5.3) and (5.8). Solving for D from these
o
expressions gives

Do =

100

= 2.18 x 10

-5

2
m /s

The value of the diffusion flux at 1273 K may be computed using these same two equations as
follows:

J =

= 1.21 x 10

-8

2
kg/m -s

5.26 To solve this problem it is necessary to employ Equation (5.7) which takes on the form
D900t 900 = DT t T
At 900C, and using the data from Table 5.2

D900 = 2.3 x 10

-5

2
m /s

= 5.9 x 10
Thus,
5.9 x 10

-12

-12

2
m /s

m /s

And
D = 4.43 x 10

-11

2
m /s

Solving for T from Equation (5.9a)

T =

101

= 1353 K = 1080C
5.27 (a) We are asked to calculate the diffusion coefficient for Cu in Al at 500C. Using the data in Table
5.2

D = D o exp

= 6.5 x 10

-5

2
m /s

= 4.15 x 10

-14

2
m /s

(b) This portion of the problem calls for the time required at 600C to produce the same diffusion
result as for 10 h at 500C. Equation (5.7) is employed as
D500t 500 = D600t600

Now, from Equation (5.8)

D600 = 6.5 x 10

-5

2
m /s

= 4.69 x 10
Thus,

t 600 =

-13

2
m /s

D500 t500
D600

4.15 x 10

102

5.28 In order to determine the temperature to which the diffusion couple must be heated so as to produce
a concentration of 3.0 wt% Ni at the 2-mm position, we must first utilize Equation (5.6b) with time t
being a constant. That is
2

x
= constant
D
Or

x2
1000

D1000

x2
T

DT

Now, solving for D utilizing Equation (5.8) in order to compute D

yields
T
1000

DT =

2
x

(2 mm) 2 2.7
x 10

= 2.21 x 10

-13

2
m /s

We now need to find the T at which D has this value. This is accomplished by rearranging Equation
(5.9a) and solving for T as
T =

Qd

R ln D
o

236, 000 J/mol

(8.31 J/mol - K) ln 2.7 x 10 -4

= 1357 K = 1084C
5.29 In order to determine the position within the diffusion couple at which the concentration of A in B is
2.5 wt%, we must employ Equation (5.6b) with t constant. That is

103

x
= constant
D
Or

x2800

D800

It is necessary to compute both D

800

and D

1000

2
x1000

D1000

using Equation (5.8), as follows:

D800 = 1.5 x 10 -4 m2 /s

= 1.22 x 10

-10

2
m /s

D1000 = 1.5 x 10 -4 m2 /s

= 1.11 x 10
Now, solving for x

1000

-9

2
m /s

yields

x1000 = x 800

= (5 mm)

D 1000
D

800

1.11 x 10

= 15.1 mm
5.30 In order to compute the diffusion time at 900C to produce a carbon concentration of 0.75 wt% at a
position 0.5 mm below the surface we must employ Equation (5.6b) with position constant; that is
Dt = constant
Or

D600t 600 = D900t900

104

In addition, it is necessary to compute both D

and D

600

900

using Equation (5.8). From Table 5.2, for

the diffusion of C in Fe, Q = 80,000 J/mol and D = 6.2 x 10

d
o

D600 = 6.2 x 10

-7

-7

900

-11

2
m /s

2
m /s

= 1.69 x 10
Now, solving for t

2
m /s. Therefore,

2
m /s

= 1.01 x 10

D900 = 6.2 x 10

-7

-10

2
m /s

gives

t 900 =

D600 t 600
D900

1.01 x 10

= 5.98 min
5.31 This problem asks us to compute the temperature at which a nonsteady-state 48 h diffusion anneal
was carried out in order to give a carbon concentration of 0.30 wt% C in FCC Fe at a position 3.5
mm below the surface. From Equation (5.5)
C x

Or

x
erf

2 Dt

105

Now it becomes necessary to, using the data in Table 5.1 and linear interpolation, to determine the
x
value of 2 Dt .
z

erf (z)

0.90

0.7970

0.8000

0.95

0.8209

From which
y = 0.9063
Thus,

x
2 Dt

= 0.9063

And since t = 48 h and x = 3.5 mm

D =

(4t)(0.9063 )2

3.5 x 10

Now, in order to solve for the temperature at which D has the above value, we must employ Equation
(5.9a); solving for T yields
T =

Qd
R ln Do

From Table 5.2, D and Q for the diffusion of C in FCC Fe are 2.3 x 10
o
d
respectively. Therefore

106

-5

2
m /s and 148,000 J/mol,

T =

148, 000 J/mol

(8.31 J/mol - K) ln 2.3 x 10 -5

= 1283 K = 1010C

Design Problems
5.D1 This problem calls for us to ascertain whether or not a hydrogen-nitrogen gas mixture may be
enriched with respect to hydrogen partial pressure by allowing the gases to diffuse through an iron
sheet at an elevated temperature. If this is possible, the temperature and sheet thickness are to be
specified; if such is not possible, then we are to state the reasons why. Since this situation involves
steady-state diffusion, we employ Fick's first law, Equation (5.3). Inasmuch as the partial pressures
on the high-pressure side of the sheet are the same, and the pressure of hydrogen on the low
pressure side is five times that of nitrogen, and concentrations are proportional to the square root of
the partial pressure, the diffusion flux of hydrogen J is the square root of 5 times the diffusion flux of
H

nitrogen J --i.e.
N

JH

Thus, equating the Fick's law expressions incorporating the given equations for the diffusion
coefficients and concentrations in terms of partial pressures leads to the following
J

2.5 x 10

1
x
x

5 JN

5
x
x

3
2.75 x 10

107

The x's cancel out, which means that the process is independent of sheet thickness. Now solving
the above expression for the absolute temperature T gives
T = 3467 K
which value is extremely high (surely above the vaporization point of iron). Thus, such a diffusion
process is not possible.
5.D2 This problem calls for us to ascertain whether or not an A -B gas mixture may be enriched with
2

respect to the A partial pressure by allowing the gases to diffuse through a metal sheet at an
elevated temperature. If this is possible, the temperature and sheet thickness are to be specified; if
such is not possible, then we are to state the reasons why. Since this situation involves steady-state
diffusion, we employ Fick's first law, Equation (5.3). Inasmuch as the partial pressures on the highpressure side of the sheet are the same, and the pressure of A on the low pressure side is 2.5 times
2

that of B , and concentrations are proportional to the square root of the partial pressure, the diffusion
2

flux of A, J , is the square root of 2.5 times the diffusion flux of nitrogen J --i.e.
A

JA =

2.5 JB

Thus, equating the Fick's law expressions incorporating the given equations for the diffusion
coefficients and concentrations in terms of partial pressures leads to the following
J

=
3
1.5 x 10

2.5 JB

2.5

108

2.0 x 10

The x's cancel out, which means that the process is independent of sheet thickness. Now solving
the above expression for the absolute temperature T gives
T = 568 K (295C)
5.D3 This is a nonsteady-state diffusion situation; thus, it is necessary to employ Equation (5.5), utilizing
the following values for the concentration parameters:
C = 0.0025 wt% N
o
C = 0.45 wt% N
s
C = 0.12 wt% N
x
Therefore

C x

= 0.2626 = 1

And thus

1

Using linear interpolation and the data presented in Table 5.1

z

erf (z)

0.7500

0.7112

0.7374

0.8000

0.7421

109

0.7374

From which

y =

x
2 Dt

= 0.7924

The problem stipulates that x = 0.45 mm = 4.5 x 10

-4

m. Therefore

4.5 x 10

Which leads to

Dt = 8.06 x 10

-8

2
m

Furthermore, the diffusion coefficient depends on temperature according to Equation (5.8); and, as
-7 2
stipulated in the problem, D = 3 x 10 m /s and Q = 76,150 J/mol. Hence
o
d

Dt = Doexp

3.0 x 10

-7

m /s

And solving for the time t

t (in s) =

0.269

exp

Thus, the required diffusion time may be computed for some specified temperature (in K). Below are
tabulated t values for three different temperatures that lie within the range stipulated in the problem.

_________________________________
Temperature

Time
110

(C)

_________________________________
500

37,900

10.5

550

18,400

5.1

600

9,700

2.7

__________________________________
5.D4 This is a nonsteady-state diffusion situation; thus, it is necessary to employ Equation (5.5), utilizing
the following values for the following parameters:
C = 0.15 wt% C
o
1.2 wt% C C 1.4 wt% C
s
C = 0.75 wt% C
x
x = 0.65 mm
Let us begin by assuming a specific value for the surface concentration within the specified range
say 1.2 wt% C. Therefore

C x

= 0.5714 = 1

And thus

1

Using linear interpolation and the data presented in Table 5.1

z

erf (z)

0.4000

0.4284

0.4286

0.4500

0.4755

111

0.4286

From which

y =

x
2 Dt

= 0.4002

The problem stipulates that x = 0.65 mm = 6.5 x 10

-4

m. Therefore

6.5 x 10

Which leads to

Dt = 6.59 x 10

-7

2
m

Furthermore, the diffusion coefficient depends on temperature according to Equation (5.8); and, as
-5 2
noted in Design Example 5.1, D = 2.3 x 10 m /s and Q = 148,000 J/mol. Hence
o
d

Dt = Doexp

2.3 x 10

-5

m /s

And solving for the time t

t (in s) =

2.86 x 10

Thus, the required diffusion time may be computed for some specified temperature (in K). Below are
tabulated t values for three different temperatures that lie within the range stipulated in the problem.

_________________________________
112

Temperature
(C)

Time
s

_________________________________
1000

34,100

9.5

1100

12,300

3.4

1200

5,100

1.4

__________________________________
Now, let us repeat the above procedure for two other values of the surface concentration, say 1.3 wt
% C and 1.4 wt% C. Below is a tabulation of the results, again using temperatures of 1000 C,
1100C, and 1200C.

C
s
(wt% C)

1.3

1.4

Temperature

Time

(C)

1000

26,700

7.4

1100

9,600

2.7

1200

4,000

1.1

1000

21,100

6.1

1100

7,900

2.2

1200

1,500

0.9

113