Sie sind auf Seite 1von 16

Diffusion in solids I

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
Keywords : Laws of diffusion, nature of concentration profile in infinite and semi infinite diffusion couples, diffusivity,
diffusion paths, effect of temperature and crystal structure on diffusivity, homogenization, hot forging, uphill diffusion,
chemical potential, activity

Introduction

Three modules of this course have been devoted to introduce the concept of diffusion in solids. Atoms &
molecules in solid, liquid and gas are never stationary. They keep moving. In a liquid or a gas one could
see or feel that this happens. If you leave a lighted candle in the corner of a room you do see the smoke
coming out and spread all over the room. Similarly if you leave some aromatic substance (naphthalene
ball) in a dish you could feel the aroma even from a distance. This happens because the molecules /
particles responsible for the smoke or the aroma are able to move around. Such a movement in the
absence of any external force is called diffusion. This takes place in solids as well. We shall look at the
phenomenological aspect and learn about the laws governing diffusion in these modules. We shall also
learn about experimental techniques based on diffusion couples that helps understand the mechanisms
of diffusion.

What is diffusion?

Diffusion is the movement of any species in a medium along a given direction. It takes place in solid
liquid or gas. It is associated with the movement of atoms / molecules of the medium and the species
present in it. It occurs even in the absence of any external force. In liquid or gas such movements can be
seen or felt easily. Slide 1 suggests a simple experiment to demonstrate the process of diffusion in a
liquid.

Slide 1: Take a glass of water with some soluble ink or


pigment at the bottom as shown in the sketch on the
extreme left. The graph beside it shows the concentration of
ink along the x axis & the height along the y axis. Note in the
beginning only the extreme bottom appears red. With time
the color keeps changing. The glass in the middle shows its
appearance after a short time. The graph beside it shows
how the concentration changes with height. The top is still
color less. The color becomes uniform after a long time.

This illustrates a case of mixing in absence of any external force. If we stir the liquid (or heat it) the
mixing will take place much faster. Mixing in the absence of any external force does occur even in solid.

Laws of diffusion:

Diffusion is the process by which a species moves in a given direction. The rate at which it moves can be
2
measured in terms of the number of species / unit area / unit time. This is termed as flux (J). Higher the
difference in the concentration of the species in neighboring region higher is the flux. The concentration
(c) is defined as the number of species in a unit volume of the medium (or matrix). The diffusion of

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
species (or material) is governed by Fick’s first law. This states that the rate of flow of material is
proportional to its concentration gradient and it occurs down the gradient. In the case of a
unidirectional flow this is expressed as:

J = —D dc (1)
ds

D is known as diffusivity. Often J is expressed as number/ (cm2 sec) and concentration is measured as
number /cm3, therefore the dimension of D is cm2/sec. The corresponding SI unit is m2/sec.

Figure 1 illustrates how concentration varies with distance.


C Note that the concentration of the species at any point does
not change with time. This is why the concentration profile of
a species in a medium is linear. This means that the
concentration gradient and the flux J remain constant.
x

Derivation of Fick’s first law:

Atoms or molecules in a medium are not stationary. They keep moving randomly in all possible
directions. Using this it is possible to derive the first law of diffusion. Let the concentration of a
particular species in a medium be c(x) at a distance x along x axis. Its concentration at x+x is c(x+x)
and its concentration at x‐x is c(x‐x). Let us consider a 2 dimensional case where atoms can move in
all the four directions. Therefore the chance that the species moves along the direction x is ¼.

c(x‐x) c(x) c(x+x) Number of species moving x along x in time t:


x‐x x x+x
y From left to right = 1 Os c(x— ∆x)
4 Ot

x From right to left = 1 Os c(x+ ∆x)


4 Ot

Fig. 2 Concentrations of species at three locations normal to the x axis in a medium are given in
the sketch. The expressions on the right gives the rate at which species keeps moving from left
to right & vice versa

From the expression given in fig 1 the expression for the net flux (J) can be obtained. Note that in the
derivation the concentration term has been written in the form a Taylor series. The higher order terms
have been neglected.
2 dc
J = 1 Os {c(x— ∆x) — c(x+ ∆x)} ≅ 1 Os{c(x) — dc ∆x + c(x) — dc∆x} = —∆s (2)
4 Ot 4 Ot ds ds 2∆t ds
3
This shows that the flux is directly proportional to the concentration gradient and the diffusivity is given
by

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
∆s2
D= (3)
2∆t

This defines the mobility of the species in the medium. It is determined by the shortest distance (jump
step) it can move and the time it takes for this to happen.

Fick’s second law:

What would happen if the concentration profile at a given time t is not linear as in fig 1. Here is a
situation where the concentration gradient keeps changing with distance. According to the first law the
flux is directly proportional to the local concentration gradient. Therefore the flux J at a given time t
would be different at different locations. This is illustrated in fig 3.

Fig 3: The sketch shows the concentration of an element in a


J1 solid as a function of distance x at given time t. Flux at any
C point is directly proportional to concentration gradient. The
flux at x‐x is J1 and that at x+x is J2. J1 > J2. This shows that
J2 more number of atoms move into the space between the two
dotted lines than that leaving this space. Therefore the
x‐x x x+x concentration of atom in this zone would increase.

Let J denote the flux at a point x. The flux at the two points at x‐x and x+x are given by the
J1 = J — 6J ∆x and J2 = J + 6J ∆x respectively. The difference between the two gives the amount of
6s 6s
the species that accumulate in this space over a small time t. Since the volume of the space between
the planes shown in fig 3 is 2x the rate of change of the concentration during the time, t, is given by

∆C
=
J1–J2
=—
6J (4)
∆t 2∆s 6s

On substitution of equation 1 and making the time interval extremely small the expression for the
change in concentration could be written as
6c
= 6 (D 6c)
(5)
6t 6s 6s

This shows that the rate of change of concentration at a given point with time is proportional to the rate
of change of concentration gradient and the diffusivity with the distance. It is known as the Fick’s second
law of diffusion. If the diffusivity is independent of concentration the law can be described as follows:

6c
=D
62c (6)
6t 6s2

The equation 6 describes the rate of change of concentration due to diffusion along x axis only. However
4
in a solid such a process can take place along any direction. Therefore we need a more general form of
the law to describe diffusion of species in 3‐D. Let us first look at the nature of diffusivity. Is this a scalar,
vector or something else? If you again go back to Fick’s first law you find that it provides a relation

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
between two vectors. Both flux J and concentration gradient are vectors. You may recall that stress too
relates two vectors; force and area. We know that stress () is a tensor of rank 2. The relation between
the force (Fi) and the area (Aj) in tensor notion is given by Fi = oijAj where the use of repeated suffix
denotes summation. Therefore diffusivity too is a tensor of rank 2. Following the above analogy flux due
6c
to diffusion along a given direction may be written as Ji = —Dij 6s . Note that concentration is a scalar.
j

Its nature is something similar to that of temperature or potential. You need two suffixes to denote
diffusivity. Depending on the material it may have different values along different directions.

Solution of Fick’s second law in a 1‐D case:

Let us look at a simple case of mixing in solid due to the movement of a particular species through the
lattice. Take the case of diffusion of carbon in iron. Carbon can dissolve in iron. Its concentration can be
increased through solid state diffusion. Keep a long piece of iron in a furnace at a given temperature. In
fig 4 this is represented by red color bar indicating that it is uniformly heated to a temperature T. For
simplicity we would consider a 1‐D case only. Therefore all the sides except the front face located at x=0
are covered. This ensures that carbon can only diffuse along the x axis. The concentration of carbon at
x=0 is maintained at cs for t > 0. This can be achieved by maintaining a suitable gaseous atmosphere
having the appropriate carbon potential to maintain the concentration of carbon at a desired level. The
boundary condition and the statement of the problem are illustrated in fig 4.

Fig 4: The rectangle denotes a piece of iron at a temperature T


at t = 0, c(x, t) = 0 for x Σ 0 in a furnace. The graph shows that the carbon content at x = 0
c is cs whereas at x > 0 it is 0. This defines the initial condition.
cs at t ≤ 0, c(x, t) = cc for x = 0
The carbon content at the exposed face is maintained at this
level at all times. The equation 6 is to be solved to get c(x,t).
x

There are standard methods to solve equation 6 for specified boundary conditions. If diffusivity is a
function of composition such equations are solved numerically. For the case described in fig 4 if D is
assumed to be independent of concentration the solution is given by the following equation.
c–c0
= 1 — erf ( s
) (7)
cc–c0 2√Dt

Where erf() denotes error function. It is given by:


z
erf(z) = 2 f exp(—y2) dy (8)
√n 0

There are standard tables to evaluate such functions. Most electronic spread sheets have this as a built
in function. Figure 5 gives a typical error function plot obtained using Excel. Note that the initial
concentration of carbon in equation 7 has been entered as c0. In this specific case it is zero. Therefore on
5
direct substitution of the initial condition in equation 7 it is possible to show that at t = 0 c = cs at x = 0 as
t approaches infinity c would approach cs at all values of x.

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
It is therefore possible to generate concentration profiles at any time t as a function of distance for a
given value of D. Figure 7 presents a set of such profiles for different values of time. In this case D has
been taken as 10‐4 cm2/s for an arbitrary species and times in the fig 7 are in hours. Note that at shorter
times the profile is stiff at lower values of x and it becomes asymptotic at higher values of x. As time
approaches infinity the concentration tends to become uniform all through the section. There is a close
similarity between the error function and standard normal distribution. This is why it is possible to
simulate diffusion as a random walk process.

0.8

0.6
erf(z)

0.4

0.2

0
0 0.5 1 1.5 2 2.5
z

Fig 5: A typical error function plot obtained from excel spread sheet.
1.20

1.00

0.80
1
0.60
C

2
0.40
5
0.20 1000
0.00
0.00 0.50 1.00 1.50 2.00 2.50
x

Fig 6: Using equation 7 the concentration of the species has been plotted as a function of distance for
different values of time.

Diffusion couple consisting of two semi‐infinite bars of metals joined at an interface:


6
Diffusion is a process of mixing. It is associated with movement of atoms in the lattice. The difference in
concentration of a given species in the matrix is the driving force. Let us consider two alloys made of two
metals (say Cu & Ni) having different compositions joined at an interface as shown in fig 7 where c

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
denotes concentration of Ni on either sides of the interface as function of distance. The alloy on the left
has lower concentration of Ni than that on the right (c1 < c2). The alloy on the left has higher
concentration of Cu than that on the right. The mobility (diffusivity) of atoms is a strong function of
temperature. If the couple is kept at room temperature the movement of atoms would be too slow to
detect any change in composition. If you heat it to a high temperature you would find that Cu atoms
would diffuse from the left to the right whereas the Ni atoms would diffuse from the right to the left.
The solution of equation 6 under this condition is given the following expression:

c–c1 = 1 {1 + erf ( s )} (9)


c2–c1 2 2√Dt

c c
c2 c2

at t = 0 at any
c1 c1 time t

x<0 0 x >0 x<0 0 x>0

Fig 7: Sketches showing how the concentration of a


c given species in a diffusion couple consisting of two
different alloys changes with time. The sketch at the
top left defines initial condition; the one at the top
(c1+c2)/2 right gives the concentration profile at any given time
at t = ∞ t and the one at the bottom left gives concentration
profile at time equal to infinity.
x<0 0 x >0
If follows from equation 9 that as t approaches infinity erf() becomes zero. Therefore c becomes
(c1+c2)/2 at all values of x. In other words it becomes a homogeneous alloy.

Self diffusion:

Does diffusion take place in pure metal? In this case there is no concentration gradient. In other words
there is no driving force for diffusion to take place. However atoms in solid are not stationary and there
are vacant sites in the lattice. Atoms in solids keep moving in the lattice by exchanging positions with
vacancies. This is illustrated in slide 2.

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
Slide 2: The sketch on the left illustrates how
Self diffusion
atoms are arranged in the lattice along with
•Does atoms in pure metal move about?
vacant sites. The line denotes an imaginary plane.
•Is there a way to find out?
The arrows represent the direction of movement
No. atoms moving towards right
= No. atoms moving towards left of atoms. At any instant the number moving to
Net flux = 0 the right is same the number moving to the left.
Therefore the net flux is zero. Is there a way to
Use radioactive tracer D*=Dt establish that diffusion takes place even in pure
Isotope has same electronic metals? Is it possible to find its diffusivity? The
structure but different mass.
Distinguishable. sketch on the right suggests that by adding tracer
atoms it may be possible to do so.
Slide 2 suggests that that the self diffusion coefficient can be estimated by adding radio isotopes at one
end of the solid and monitoring how these atoms move into the solid. Isotopes are atoms having
identical electronic structure but different mass. Often the mass differs by one atomic mass unit. Such
atoms are unstable they try to get back to the more stable structure by emission of radiations. This can
be detected by suitable measuring devices like counters. If the concentration of tracer atoms can be
obtained as a function of time & distance; the diffusivity of the tracer can be estimated. Since the size
and the mass of the tracer atom and the base atom are nearly the same it may be assumed that the
diffusivity of the two species would be the same.

Temperature dependence of self diffusion coefficient:

We have just seen how the self diffusion coefficient can be estimated using radioactive tracer.
Diffusivity (D) of a given species in a specific medium can also be obtained from the concentration
profile along a diffusion couple. Diffusion of species is associated with the mobility of atoms. This
increases with increasing temperature (T). Therefore it is expected to be a strong function of
temperature. The temperature dependence of diffusivity is often represented as follows:

D = D0exp (— Q ) (10)
R
T
Where R is the universal gas constant, Q is the activation energy for diffusion and D0 is a constant. If D is
known at several temperatures Q & D0 can be obtained from the slope and intercepts of the plot given
in fig 8.

D0
Slope = Q/R Fig 8: Illustrates temperature dependence of diffusivity.
ln(D) If follows from equation 10 that ln(D) = ln(D0) — Q .
R
Therefore intercept = ln(D0) and slope = ‐Q/R. T

1/T

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
Diffusion paths in solids:

Diffusion of species in solids also depends on the path that it follows. In a solid we can think of three
distinct paths. Metals are made of several crystals that meet along grain boundaries. If a species has to
move through this it could either move through the grain, the grain boundary or the top surface. This is
illustrated in slide 3. The diffusivity through the grain is denoted by Dg. This is often known as bulk
diffusion coefficient. The diffusivity through the grain boundary is denoted by Dgb. The space between
atoms because of irregular arrangement is more at the grain boundary than that within the grains. This
is why the mobility of atoms through the grain boundary is expected to be higher than that through the
grain. The same logic can be extended to the exposed top surface. There is enough space to
accommodate extra atoms at the free surface if required. Therefore the mobility of the atoms along the
free surface (Ds) is much higher. The relation between the three could be described as Ds > Dgb > Dg.

Slide 3: Shows three distinct paths in a solid through


Diffusion paths in solids
which an atom of a given species could diffuse. The
surface diffusivity of the species through these paths is
grain Grain denoted as Ds (surface diffusion), Dgb (grain
boundary
boundary diffusion) and Dg (diffusion through the
grain). These are strong functions of temperature.
Ds
D Ds > Dgb > Dg The temperature dependence is very much similar
Dgb
Dg to the expression given in equation 10. However the
1/T magnitudes of Q & D0 are likely to be different.

The slide 3 also illustrates the temperature dependence of the diffusivity through three different paths.
The activation energy of surface diffusion is likely to be the lowest and that for the grain is the highest.
Mathematically this is denoted as Qg > Qgb > Qs.

Diffusion as a random walk process:

The process of diffusion is governed by the movement of atoms. In solids where atoms are closely
packed such movements will be difficult in the absence of vacant sites. At a given temperature several
lattice sites are vacant. If there are several sites around an atom how would an atom decide where to
move? An obvious option could be a random selection. Slide 4 illustrates the difference between a
normal and a random walk.

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
Diffusion as random walk
Slide 4: If a species moves through the lattice at
a velocity v, the distance between its initial and
Distance cover in time t during final position after time t is equal to vt, provided
normal walk along a specific
direction= velocity x time = v t it continues to move in the same direction. Note
that the long red line with an arrow head denote
the distance covered in this case. However if the
Distance covered in time t
during random walk along a direction keeps changing randomly the distance
specific direction << v t between the initial and final position is much
n
less. This is shown with the help of a shorter red
Rn  r1  r2  rn   ri line with an arrow head.
i1

The distance between the initial and the final position is best represented as a vector sum of each step
(ri̅ ). The final location (R̄n ) after n steps of movement is given by:

R̄n = r1 + r2 + … . +rn = ∑ ni i=1r (11)

The dot product of a vector with itself gives the square of its magnitude. This is given by
R̄ 2 = ∑n r̅ 2 + ∑n ∑n r̅ r (12)
n i=1 i i j*i i j

Each term of the second series in equation 12 denotes dot product of two vectors. If the angle between
the two is ij, and each step size is equal to  the equation 12 on simplification becomes:
R̄ 2 = nh2 + h2 ∑n ∑n cos 8 (13)
n i i*j ij

Since the direction of motion is random the magnitude of cos could be both and negative. The sum
total value is therefore likely to be zero. Thus the average distance between the initial and final location
in this case is h√n as against n in the case of normal walk. R̄n represents average root mean square
distance. In equation 3, x/t denotes the average velocity (v) of the diffusing species and x is the
average step size which equivalent to . The total distance covered by the species is n. Therefore the
time it takes to cover this distance is n/v. Thus by a little algebraic simplification you get the following
relations:

D=
∆s2
=
ßv
=
nß2 or h√n = √2Dt (14)
2∆t 2 2t

This shows that average random walk distance is equal to √2Dt.

Effect of crystal structure on diffusivity:

10 As we go through the course we would see how the composition of solid can be altered by allowing
certain atoms to move into it by diffusion. Often it becomes necessary to increase the carbon content at
the surface of steel which is mostly made of iron. Iron exists in two different crystalline forms. At room
temperature it is BCC whereas above 910°C it becomes FCC. The solubility of carbon in FCC state is

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
higher than that of BCC. If we want the steel to dissolve substantial amount of carbon we must take it to
beyond 910°C. However carbon can also diffuse in its BCC state. The rate of carbon pick up would
depend on its diffusivity. The BCC form of iron is known as ferrite and often represented as . The FCC
form of iron is known as austenite. It is represented as . The diffusivities of carbon in these two forms
of iron are as follows:

Dα = 2 × 10–6exp (— 84400) m2/s (15)


c RT

Dy = 2 × 10–5exp (— 143000) m2/s (16)


c RT

If you estimate these over a range of temperatures (600°‐1000°C) and plot the data against reciprocal of
temperature in K you get a plot as shown in 9.

1.00E‐09

1.00E‐10
D m2/s

Ferrite
1.00E‐11 Austenite

1.00E‐12
6.00 7.00 8.00 9.00 10.00 11.00 12.00
10000/T, K‐1

Fig 9: Diffusivity of carbon in two forms of iron ferrite and austenite as function of temperature

Austenite is a close packed structure in comparison to ferrite. Therefore diffusivity of carbon in


austenite is lower than that in ferrite. The solubility of carbon in austenite is much more than that of
ferrite. Therefore in order to have uniform distribution of carbon in steel it has to be heated to
austenitic state. The diffusivity of carbon in austenite being very low the time needed homogenization
of cast steel can be very long.

How to reduce time needed for homogenization:

Solidification of metals and alloys always results in segregation. This is because the only way a species in
a solid can move is by diffusion which is a slow process. It means cast structure is rarely homogeneous.
11 In order to make the solid homogeneous it is often held at a high temperature for long hours. The
effective diffusion distance is approximately equal to √Dt . A simple calculation would show that the
time needed for homogenization even if the distance over which heterogeneity exists is of the order of a

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
few mm could be very long. Hot forging is a process that combines diffusion with mechanical working.
The combined effects of diffusion and applied stress having a churning effect on the solid can help cut
down the time needed for homogenization significantly. Solidification of metals and alloys always results
in segregation. This is because the only way a species in a solid can move is by diffusion which is a slow
process. It means cast structure is rarely homogeneous. In order to make the solid homogeneous it is
often held at a high temperature for long hours. The effective diffusion distance is approximately equal
to √Dt . A simple calculation would show that the time needed for homogenization even if the distance
over which heterogeneity exists is of the order of a few mm could be very long. Hot forging is a process
that combines diffusion with mechanical working. The combined effects of diffusion and applied stress
having a churning effect on the solid can help cut down the time needed for homogenization
significantly. Applications that demand homogeneous alloys prefer to use components that have been
produced by hot upset forging.

Uphill diffusion:

Diffusion in solid as described above has been visualized as a process of mixing. The concentration
gradient has been assumed to be its driving force. The mass flow takes place down the concentration
gradient. This may true for solid solutions having a single minimum in its free energy composition plot as
shown in fig 11. However there are several alloy systems where diffusion may take place against the
concentration gradient. Such a system has free energy composition plots having multiple minima as
shown in fig 12. This is known as uphill diffusion. This is because the true driving force for diffusion is the
chemical potential or the partial molar free energy and not the concentration. The molar free energy (G)
of a binary solid solution consisting of two metals A & B is given by:
G = NÆG0 + NBG0 + RT[NÆln(aÆ) + NBln(aB)] = NÆµÆ + NBµB
Æ B

NA & NB denote atom fraction A & B, G0 & G0denote the free energies of pure A & B, R is the universal
Æ B
gas constant, T is the temperature in °Kelvin, A & B denote the chemical potentials of A & B and aA &
aB denote the activities (effective concentrations) of A & B. The two activities are given by

aÆ = yÆNÆ : Activity coefficient yÆ = 1 in an ideal solid solution


aB = yBNB : Activity coefficient yB = 1 in an ideal solid solution

Figure 10 shows a typical free energy composition diagram of a binary alloy at a given temperature. The
intercept of the tangent at a point G1 with the vertical axis at A denotes µ1Ædenotes the partial molar
free energy of A in alloy 1. It is also known as its chemical potential. The intercept of the same tangent
with the axis B denotes the chemical potential of B in alloy 1 denoted as µ1B. Note the magnitudes of the
chemical potentials of A in 2 (µ1) and B in 2 (µ1). Figure 10 includes the sketch of a diffusion couple
Æ Æ
made of two alloys 1 & 2 having different compositions. Alloy 1 is rich in A whereas alloy 2 is rich in B. In
this case µ1 Σ µ2 therefore A diffuses from 1 to 2. Here N1 Σ N2as well. It represents a case where
Æ Æ Æ Æ
diffusion takes place down the concentration gradient.
12
Figure 11 shows a typical free energy composition diagram of a binary alloy at a given temperature. It
has two minima. The intercept of the tangent at a point G1 with the vertical axis at A denotes µ1Ædenotes
the partial molar free energy of A in alloy 1. The intercept of the same tangent with the axis B denotes

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
the chemical potential of B in alloy 1 denoted as µ1B. Note the magnitudes of the chemical potentials of
A in 2 (µ1Æ) and B in 2 (µ1).ÆFigure 11 includes the sketch of a diffusion couple made of two alloys21 & 2
having different compositions. Alloy 1 is rich in A whereas alloy 2 is rich in B. In this case µ Σ µ1
Æ Æ
therefore A diffuses from alloy 2 to 1 although N1 Σ N2. It represents a case where diffusion takes
Æ Æ
place against the concentration gradient.

1 A 2
B Fig 10: The sketch at the top represents a
diffusion couple consisting of two alloys 1 & 2.
Alloy 1 is rich in A and alloy 2 is rich in B. G
represents the free energy composition plot. G1
G is the free energy of 1 and G2 is the free energy
of 2. Gm is the free energy of the most stable
2 composition. The vertical arrow head ending at
µB
1
µÆ G1 G2 this point is a measure of the driving force for
diffusion. Note the points of intersection of the
tangent at G1 & G2 with the vertical axis at A.
Gm µ1 1 2
B Since µÆ Σ µÆ A diffuses from 1 to 2. Diffusion

2 direction is shown with the help of an arrow.


µÆ
µ2 Σ µ1 : Hence B diffuses as shown from alloy
B B
2 to 1. This is a case of downhill diffusion
A 1 2 B

µ 1B Fig 11: The sketch at the top represents a


1 A rich B rich 2 diffusion couple consisting of two alloys 1 & 2.
2 A Alloy 1 is rich in A and alloy 2 is rich in B. The
µÆ B sketch at the bottom gives free energy
composition plot G. G1 is the free energy of 1
and G2 is the free energy of 2. Ga denotes the
combined free energy of 1 & 2. Gm is the free
G energy of the most stable composition. The
vertical arrow head ending at this point is a
G1 Ga measure of the driving force for diffusion.
G2
Note the points of intersection of the tangent
Gm
at G1 & G2 with the vertical axis at A. Since
µ2 Σ µ1 A diffuses from 2 to 1. Diffusion
Æ Æ
µ1
Æ direction is shown with the help of an arrow.
µ2 µ1 Σ µ2 : Hence B diffuses as shown from
13 B B B
A 1 2 B alloy 1 to 2. This is a case of uphill diffusion.

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
This suggests that it is the chemical potential gradient that determines the direction of flow of species.
Therefore the more general expression for Fick’s law should be expressed in terms of either chemical
potential or activity gradient.

The best known example of uphill diffusion was first reported by L S Darken in Trans AIME Vol. 180
(1949) p430‐438. A diffusion couple was made by welding two pieces of steel one having 3.5wt % Si &
0.49wt % C and the other having 0.05wt % Si & 0.45wt % C. This was kept to 1050°C for 13 days. At such
a high temperature the most stable form of iron is austenite having FCC structure. The solubility of
carbon in austenite is much higher than the amount of carbon present in the two steel. Figure 12 shows
with the help of a schematic diagram carbon concentration profile before and after higher temperature
thermal exposure. No doubt carbon is expected to diffuse from the steel on the left having 0.49%C to
that on the right having 0.45%C. However % carbon near the interface on the left was found to have
gone down to 0.33% whereas that on the right of the interface had gone up to 0.59%. Clearly this is a
case that illustrates diffusion of a species in a direction against the concentration gradient. This unusual
phenomenon was attributed to the presence of Si that alters the activity (or the chemical potential) of C
in iron.

3.8Si 0.49C 0.05Si 0.45C

0.55%C
Time = 0 Time = 13days

0.49%C

0.45%C
Time = 13days
0.33%C Time = 0

Fig 12: The sketch at the top shows a schematic representation of a diffusion couple made of two
steel having different amounts of C and Si. The sketch at the bottom shows the initial (t = 0) and
the final (t = 13days at 1050°C) concentration of C in the two steel.

Summary

In this module we have learnt about a solid state mixing process called diffusion. This is governed by
Fick’s laws of diffusion. The first law describes steady a state process when the concentration of a given
species at a point does not change with time. In this case the concentration gradient remains constant
14 at all points. If the concentration gradient is a function of distance then the concentration at a point
would keep changing with time. This is given by Fick’s second law. This is expressed in the form of a
differential equation. The nature of solution of such an equation in 1D has been discussed without going
into its derivation. Attempts have been made to give an insight into its mechanism. It can be simulated
NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
as a random walk process. The average random walk distance has a direct correlation with diffusion
distance. We also looked at the effect of temperature on diffusivity. Using an illustration it has been
shown that diffusion depends on temperature as well as crystal structure. Although Fick’s law is often
written in terms of concentration gradient there are cases where diffusion may takes place against the
concentration gradient. This is known as uphill diffusion. This suggests that it is more appropriate to
define this in terms of chemical potential.

Exercise:

1. If iron is kept at 1200˚K in a carburizing atmosphere for 8hrs to obtain a carbon


concentration of 0.75 at a depth of 0.5mm. Find the time it would take to reach same
carbon concentration at depth of 7.5mm at 1250˚K. (Given D0 = 0.2x10‐4 m2/s & Q =
143kJ/mole/˚K)

2. A steel containing 0.2 % carbon was heated to 950˚K for 15 hours. Find the depth of layer in
which there is no carbide. Assume that steel consists of ferrite and carbide. The solubility of
carbon in ferrite at this temperature is 0.015% and % C at the surface is negligible. (Given D0
= 2x10‐6 m2/s & Q = 84.4kJ/mole/˚K)

3. The concentration of carbon on the surface of iron is maintained at 1.00% at 1175˚K for
2hours. Estimate the depth at which % C would be 0.5%. Use the diffusivity values given in
question1. Assume initial carbon content of iron to be negligible.

Answer:

1. If iron is kept at high temperature in an environment having high carbon potential it diffuses
s Dt
into iron. The depth of carburization (x) is proportional to√Dt. Therefore 1 = J 1 1 Since
s2 D2t2
Q s1 2 t1 Q 1 1 0.50 2 8 143000 1

D = D0exp (— RT) ( s)2 = exp { ( — )} Therefore ( ) = exp{ ( —


t2 R T2 T1 0.75 t 8.33 1250
11200=10.14 hours

2. If initial carbon in steel is Ci & soluble carbon in ferrite is C amount carbon to removed / unit
cross section through a distance dx = (Ci‐C)dx = flux of carbon atom in time dt = ‐Jdt =
dC –0
D dC dt Assuming carbon concentration at surface as 0 = Cα Thus (Ci — C0)dx = D Cα dt
ds 2DtCα Q 84400 ds s s
or; x = J D = D exp (— ) = 2 × 10–6exp {— }= 4.55 × 10–11m2/s Now C =
(Ci–C0) 0 RT i
8.31×950
2×4.55×10—11 ×0.015×15×3600
0.2 & Ca = 0.015 Therefore x = J =0.00063m = 0.63mm
(0.2–0.015)

15
3. Carbon content at surface Cs = 1.00, Initial carbon C0 = 0.0 & carbon content at a distance x
C=0.5

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||
1.0 – 0.5 s
Cc–C = erf [ s ] Or, = erf [ s ] Or, = 0.477
Cc–C0 2√Dt 1.0 2√Dt 2√Dt

Or, x = 0.477 × 2 × J0.2 × 10–4 exp (— ) × 2 × 3600 =0.000425m =0.445mm


143000
8.31×1275

16

NPTEL Phase II : IIT Kharagpur : Prof. R. N. Ghosh, Dept of Metallurgical and Materials Engineering || |
||

Das könnte Ihnen auch gefallen