PMMA looks like a pretty good choice for large, cheap, relatively broadband optics.
It's got a fairly flat ~90% transmission for a few-mm sheet (good enough, I can make up for the loss by increasing objective diameter) from around 370-1100nm then falls off sharply into the UV and has a small dip in the IR before rising again.
So it'll cover the vast majority of my spectrometer's capacity from mid UV-A into NIR, just losing the UV-B/C on the far short end of my wavelength range (which the atmosphere doesn't transmit all that well anyway).