To consider mass and electrical transport through a material, it is helpful to have an understanding of diffusion first at a continuum level i.e. without atomic details. After this, well introduce the atomic details.
Overview of diffusion and Ficks Laws
Consider a distribution of a diffusing particle in 1D of the form shown below.
If diffusion in any direction were equally probable, you would expect that at a later time, the distribution would spread out as shown below. Well bin particles to see how that happens.
Fick thought about this idea and came up with a couple of laws expressed mathematically below. Well first consider Ficks first law. It says that the flux of particles through a point (in 1D) or an area (in 3D) is equal to the negative of the change in concentration with respect to displacement times the diffusion constant. The diffusion constant is proportional to the average displacement squared per unit time. Ceramics Gomez Page 2 out of 5
J = !D "C "x Ficks First Law C(x,t) is the concentration at position x and time t. The concentration of particles on a line is the number of particles per unit length. In three dimensions, it would be the number of particles per unit volume. D is the diffusion coefficient. It is proportional to the average square displacement for a specific time interval per unit time and hence has units of length 2 /time. All the atomic information is buried in D. J(x,t) is the diffusing particle flux. Its proportional to the change in concentration with respect to location of flux. The diffusion constant is the proportionality constant. Hence, in one dimension the flux has units of number of diffusing particles per unit time. In three dimensions, the flux has units of number of diffusing particles per unit area per unit time. This makes sense since in one dimension we speak about flux through a point where as in 3D there is flux through an area.
Lets see if Ficks first law makes sense. Lets plot the negative derivative of the concentration profile as a function of position thats almost the flux.
Notice that to the right of the initial peak, the change in concentration with respect to position change is positive i.e. particles move to the right. On the left, the change in concentration with respect to change in position is negative i.e. particles move to the left. So, the flux just shows us how the diffusion leads to a spreading out of the concentration. The magnitude is directly proportional to the change in the concentration at each point and the diffusion constant. The diffusion constant is how continuum theories take into account differences in materials.
Lets consider Ficks second law. It states that the change in concentration per unit time is equal to the negative of the change in flux per unit displacement. See that the units make sense. Plugging in Ficks first law, we obtain:
Ceramics Gomez Page 3 out of 5 !C !t = " !J !x = D ! 2 C !x 2 Ficks Second Law
For the Ficks second law, its helpful to draw the second derivative of the concentration i.e. consider how J changes with position. That is shown below.
Note that this plot tells us that the concentration decreases most as a function of time at the original maximum in concentration and increases as a function of time at the original tails of the distribution. Basically, this equation shows us that the concentration distribution broadens with time.
Weve been getting some intuition about diffusion through a plot of a solution to the diffusion equation. You might wonder, how does one find the diffusion equation. Unit analysis can help us pick the form of the concentration. Our goal will be to express the concentration as a product of a function that will carry the units and a unitless one, which will carry any remaining variable dependence. For example, we know that in 1D, the units of concentration are number of particles per unit distance. The key quantities we have to play with are total number of particles (N), the diffusion constant, the time, and the position. Notice that the units of position and time are related through the diffusion constant. More specifically, the units of position could be obtained by considering the units of (Dt) 1/ 2 . To get the units for the concentration (number of particles per unit distance), we could mix the variables in the form N (Dt) 1/ 2 but still would have no position dependence. However, what we have learned is that it might be possible to find a solution to the diffusion equation by considering a function with the correct units times a unitless function e.g. C(x, t) = N (Dt) 1/ 2 f x (Dt) 1/ 2 ! " # $ % & .
Ceramics Gomez Page 4 out of 5 Notice that the argument of the unitless function is unitless. Lets see if we can find a function that satisfies Ficks second law but putting this form into Ficks second law. When doing this consider the substitution z = x (Dt) 1/ 2 .
!C !x = N (Dt) 1/ 2 !f (z) !x = N (Dt) 1/ 2 !f (z) !z !z !x = N (Dt) 1/ 2 !f (z) !z 1 (Dt) 1/ 2 = N Dt !f (z) !z ! 2 C !x 2 = N Dt ( ) 3/ 2 ! 2 f (z) !z 2 !C !t = " 1 2 DN (Dt) 3/ 2 f z ( ) + N (Dt) 1/ 2 !f (z) !z !z !t = " 1 2 N D 1/ 2 t 3/ 2 f z ( ) + " 1 2 Dx (Dt) 3/ 2 N (Dt) 1/ 2 !f (z) !z !C !t = " 1 2 N D 1/ 2 t 3/ 2 f z ( ) + " 1 2 N D 1/ 2 t 3/ 2 z !f (z) !z = " 1 2 N D 1/ 2 t 3/ 2 f (z) + z !f (z) !z # $ % & ' ( !C !t = D ! 2 C !x 2 " 1 2 N D 1/ 2 t 3/ 2 f (z) + z !f (z) !z # $ % & ' ( = D N Dt ( ) 3/ 2 ! 2 f (z) !z 2 0 = ! 2 f (z) !z 2 + 1 2 f (z) + 1 2 z !f (z) !z 0 = d dz df dz + z 2 f ) * + , - .
Our dimensional analysis has allowed us to turn Ficks law with had two variables into a differential equation of one variable. Integrating yields:
df dz + z 2 f = constant=0
Since as the position of observation goes to infinity, the concentration and change in concentration should go to zero, hence the integration constant is zero. At this point, weve enforced a boundary condition. If our material had edges prior to infinity, this would have worked out differently. Now, continuing with our equation:
Ceramics Gomez Page 5 out of 5 df dz = ! z 2 f df f = ! z 2 dz ln f / C = !z 2 / 4 f = Ce !z 2 / 4 = Ce ! x ( Dt ) 1/ 2 " # $ % & ' 2 / 4 = Ce ! x 2 4Dt " # $ % & ' C(x, t) = N (Dt) 1/ 2 Ce ! x 2 4Dt " # $ % & '
Finally, to find out the constant C, we note that the integral over all the concentration should be the number of particles.
N (Dt) 1/ 2 Ce ! x 2 4Dt " # $ % & ' dx !( ( ) = N C (Dt) 1/ 2 e ! x 2 4Dt " # $ % & ' dx !( ( ) = 1 C (Dt) 1/ 2 * 1 / (4Dt) " # $ % & ' 1/ 2 = 1 C = (Dt) 1/ 2 4Dt* ( ) 1/ 2 = 1 4* ( ) 1/ 2 C(x, t) = N (4*Dt) 1/ 2 e ! x 2 4Dt " # $ % & '
and voila, we have the concentration as a function of position and time. Plotting it reveals that is just what we made pictures of earlier. If the initial source had a different shape or the boundary condition had been different, the resulting function would have differed as well. Plots of some functions like this are shown in Table 3.1. Greens functions provide a nice mathematical device to get these solutions in a general way. For now, its enough to know that its possible to get them. Now, that we have explored diffusion macroscopically, lets go back to think about the atomic details of diffusion.