I suggest you use image reversal (after lithography, width of opening at surface
is larger than an top of photoresist, upon sputtering or e-beam, this produces
an opening above the metal through which acetone can dissolve the PR) - do a
google search or visit a library of course. I liftoff niobium 1..5 micron thick,
3-micron lines, 7-micron spaces diffraction gratings on silicon. I do find I
need to soak in acetone for 20 minutes followed by ultrasonication for 20
seconds or so. Good luck.

Thomas