In the polysilicon micromachining arena, I think that the record is currently
held by Ming Wu at UCLA. I've been helping him make free-space optical
components (mirrors, gratings, lenses) using micro-hinged polysilicon. At least
one of his systems had an optical axis 1 mm above the surface of the chip, so
his fresnel lenses topped out at around 1.5 mm .

Based on my experiences w/ polysilicon, my guess is that the first height limit
that you'll reach will be dictated by the die size (say 1cm), but there is the
possibility of having a multiply hinged structure fold up to many times the die
width. I'll be giving my MEMS design class a homework problem soon where they
try to make the tallest structure out of a 1mm square area using the MCNC
design rules. I'll let you know if anyone breaks 1cm.

ksjp