There are a few EUV transparent materials for light source windows, including silicon and zirconium. See http://www.luxel.com/wp-content/uploads/2010/04/EUVL-Paper.p...

There are other concerns wrt contamination though, e.g. masks degrading in the EUV radiation and emitting contaminants that end up on the wafer. This is more likely to happen at higher power levels.

These and other concerns are discussed at Chris Mack's website http://life.lithoguru.com/ (he is quoted in the linked article). His series of posts on the SPIE Advanced Lithography Symposium 2016 is particularly interesting and seems to be the basis of most of the more-popular press articles on EUV in the last week.

Very interesting link. I at first wondered how such a thin film would be strong enough, then realized they intend to have both sides of the window in vacuum so there is no pressure differential, and no need for strength. (Which should have been obvious since air is opaque.)

That actually made me think of another question.

What do you make the mask out of? Those transparent materials are transparent only in incredibly thin films.

Indeed, instead or lenses the EUV machines use mirrors. The vacuum makes things more difficult as well, you can't cool much to the air.

I think they finally managed the tin contamination, especially the first mirror or so was a big issue.

Source: some friends working at ASML. It's a big employer in its region, Eindhoven, Netherlands.

Apparently losses are as high as 50% per mirror?

IIRC only 2% of the generated light eventually hits the wafer. The other 98% of that energy needs to be cooled away.

Why it has to be transparent?

There are projectors that do not use transparency but reflection. Think DLP or LCos on the digital world but with analog masks.