Вы находитесь на странице: 1из 6

Lecture 4: Diffusion: Ficks second law

Todays topics Learn how to deduce the Ficks second law, and understand the basic meaning, in comparison to the first law. Learn how to apply the second law in several practical cases, including homogenization, interdiffusion in carburization of steel, where diffusion plays dominant role.

Continued from last lecture, we will learn how to deduce the Ficks second law, and understand the meanings when applied to some practical cases. Lets consider a case like this

C j(x)

A j(x+dx) A

x x+dx

dx x

We can define the local concentration and diffusion flux (through a unit area) at position x as: c (x,t), J(x) we have [ j ( x) j ( x + dx)]tA c(x) = , Adx then, we have
c( x, t ) j =x t

J(x+dx) = J(x) +

j x x

form the first law: J = -D then, we have

dc( x) dx

2c c( x, t ) j ===D 2 x t x


This is the Ficks second law. In three-dimensional space, it can be written as: c = D 2c . t At steady (equilibrium) state, we have
c( x, t ) /dt = 0

(meaning no concentration change)

Then, solving Eq. (1) gives dc -D = J = constant --- back to the Ficks first law. dx So, Ficks first law can be considered as a specific (simplified) format of the second law when applied to a steady state.

Now, lets consider two real practical cases, and see how to solve the Ficks second law in these specific cases. Case 1. Homogenization: (non-uniform uniform) Consider a composition profile as superimposed sinusoidal variation as shown below, where the solid line represents the initial concentration profile (at t=0), and the dashed line represents the profile after time .
C t=0

0 _ C l x t=



c= c + 0 sin


with the Ficks second law, constant.

c 2c = D 2, x t

where D is teh diffusion coefficient, a

At time t,

c(t )= c + 0 sin


Where = l / D, is defined as the relaxation time.

l The largest Deviation of concentration at x = , where (t) =0 e t / 2 It is an exponential decay, the longer the wavelength (l), the longer the relaxation time (), then the slower decay. Short wavelength dies fast. Thats why shaking always helps speed up the dispersion, because it enables wide spreading (smaller l) of the stuff (like particles) you try to disperse.

Case 2. Interdiffusion (the carburization of steel):doping of steel with carbon

c CH4 CO C cs 1/2 (cs+co)

x =0


c0 0 x

Carbonization thickness is defined as (cs+c0) =


Consider a real example: carbon diffusion an austenite ( phase of steel) at 1000 C, D=410-11 m2s-1, carbonization of 0.2 mm thick layer requires a time of ca. 1000 seconds, or 17 min.

Situation a): Doping with fixed amount of dopant Consider a thin layer of B deposited onto A, through annealing at high temperature, we will be able to get the concentration profile at different times, from there then we can determine the diffusion coefficient, D

B 0

A x

Infinite integration of the Ficks second law, We have c(x,t) =

2c c =D 2 x t


/ 4 Dt


As the B diffuses into A, the total amount of B is fixed

c( x, t )dx = N = constant


x 2 / 4 Dt

dx = 2

x 2 Dt



x )=N Dt

To solve the above equation, lets define y = 2 since then, = N D

x then we have, 2 Dt

e y dy = N

e y dy =

so Eq. (2) can now be written as c(x,t) =

2 N e x / 4 Dt Dt


as determined by this diffusion kinetics equation, the concentration profile of carbon as various times will be like this
C t1 t3 > t2 > t1


t3 x

The above diffusion is one-direction (0 +). But if we extends it to two-way, from - to + (like a droplet dissolved into a solution) with dopant at x=0, then we have
t1 t2 c (x,t) = t2 > t1 = N

2 D
2 N e x / 4 Dt , x (-, ) 2 Dt

Situation b): Doping with a fixed surface concentration (e.g. carburization of steel)




Carbon concentration profile shown at different times, Carbonization thickness is defined as (cs+c0) = Dt

The solution of the Ficks second law can be obtained as follows, the surface is in contact with an infinite long reservoir of fixed concentration of Cs. For x < 0, choose a coordinate system u.
diffuse Cs du u x=0 x X

The fixed amount of dopant per area is Cs d=N, which diffuse toward right. Then using Eq. (3) above, the slab du contributes to the concentration at x is dc(x, t) = Cs du 2 / 4 Dt e Dt

So, all the slabs from x= - to x totally contribute c (x, t) =

dc( x, t ) =

2 cs e / 4Dt d Dt

Defining y = 2c s

2 Dt

, then,

c (x, t) =

2c s

x / 2 Dt

e y dy ,


e y dy 2

x / 2 Dt

e y dy ]

x / 2 Dt

e y dy ]

= cs [ 1- erf (

x )] 2 Dt 2

Where error function erf (z) =




Considering boundary conditions: c (x = 0) = cs , c (x = ) = c0, corresponding to the original concentration of carbon existing in the phase, c0 remains constant in the far bulk phase at x = . c (x, t) = cs (cs c0)erf ( x ) 2 Dt

the concentration profile shown above follows this diffusion equation. Now lets consider Interdiffusion as shown below, which represents more general cases.
c1 > c2 C c1 C1 C2
c1 + c2 2


c2 x=0 0

Solving the Ficks second law gives

c (x,t) = (

c1 + c2 c + c2 x )( 1 ) erf ( ) 2 2 2 Dt

Interdiffusion is popular between two semi-infinite specimens of different compositions c1, c2, when they are joined together and annealed, or mixed in case of two solutions (liquids). Many examples in practice fall into the case interdiffusion, including two semiconductor interface, metal-semiconductor interface, etc.