How are snowflakes formed?
We now know that snow crystals grow into snowflakes because of three processes acting on them at once. The first two are known as faceting and branching. Imagine the ice crystal building block of any flake as a miniature hexagonal ice hockey puck. Its six sides are prism facets. Its top and bottom are its basal facets. As it falls through cold, moist air, or swirls around in it heedless of gravity, free-floating water molecules from evaporated droplets adhere directly to these facets without passing back through the liquid phase. When they stick to the basal facets the puck thickens. It can quickly become taller than it is wide – a column rather than a puck.
When they stick to its prism facets it becomes wider, like a plate, and one of the mysteries of faceting is that both versions can act on one flake but they very rarely do so at the same time. This is there are snowflakes that look like pairs of train wheels on an axle. They grow first through basal faceting to create a column, then through prism faceting to create the wheels, with an abrupt switch from one to the other than no one can quite explain.
For all the oddness of train-wheel shapes, snowflakes generally would be dull if they were shaped only by faceting. They would be small and granular and lie heavy on the ground, like the rough little crystals that pile up next to snow cannons on denuded ski slopes.
What makes snowflakes beautiful is branching. In a crystal’s journey through the atmosphere all the parts of its surface are in competition with each other for free-floating water molecules. In this “competition”, to stick out into the air is to have the edge, so when the prism facets are growing, the corners between them grow faster than the facets themselves. As the corners begin to stick out they catch even more molecules relative to the self-effacing facets, and so they grow faster – a positive feedback loop that creates the branches that turn crystals into stars and snow into something light, fluffy and miraculous.