This seems a bit advanced. I saw this forum where that exact issue is being discussed:http://www.physicsforums.com/showthread.php?t=191812
The discussion sounds way above the 6-8 grade level. What are your calculus and programming skills? Have you ever written a program that solves differential equations? What approach did you have in mind? Here is a free/open source software package that might help:http://freshmeat.net/search?q=diffusion ... mit=Search
although it may be a challenge to figure out (I did not look at it closely) and may be focused on something you don't really care about directly. You can also search with keyword sets such as:
"diffusion equation" program java
"diffusion equation" program python
"diffusion equation" program c++
Alternatively, you could take an entirely different approach and write a program that moves balls or circles around on a 2 dimensional surface at random, e.g. Brownian motion, with or without collision detection and see if it simulates diffusion in a realistic manner. You'll be able to vary how much each ball moves during each time stamp, how many time steps to simulate, output the simulation to an animation, and probably a few other things. If you have any experimental data you want to try and reproduce, you would in effect be trying to discover your own scientific model of diffusion.
Please let us know what you think.