I also didn't know why they used silicon. Silicon is the second most abundant element in the Earth's crust (after oxygen), and yet was unknown until the 19th century because it bonds so tightly in chemical compounds. It seems silicon was used for this because we know so much about its crystal structure.
Of course, after watching the video I went to find out more and found this article by the lab that produced the spheres. Interesting, btu I must take offense at their claim:
CSIRO’s Australian Centre for Precision Optics is the only place in the world capable of fabricating round objects with the accuracy required for the Avogadro Project.and
The best sphere the ACPO team has ever made had a total out-of-roundness of 35 nanometres. That is, the diameter varies by an average of only 35 millionths of a millimetre, making it probably the roundest object in the world.Now, as an amateur telescope maker I have to say that's good, but not great. I have made spheres accurate to 5nm, so these are certainly not the roundest objects in the world. Here is a graph of a mirror made by a friend of mine showing the surface error in nm (I'm assuming he doesn't mind me using an image displaying his optical fabrication prowess).
I was also curious as to how they tested a convex sphere to that accuracy. It is easy to test a concave sphere, much harder to test a convex sphere. The video indicates that they used a laser spinning around the sphere. But that test would depend on the accuracy of the bearing the laser spun on. I guess that would have to be averaged out over many tests.
Either way, this is a fascinating topic, that bears witness to the beauty and complexity of even the simplest thing in nature.