I think the standard derivation of the Gaussian integral uses the same technique. It's between equations (2) and (3) of

http://mathworld.wolfram.com/GaussianIntegral.html

You make the single integral into a double integral, and then the double integral turns out to be easy to do using polar coordinates.