<azonenberg>
My tentative plan calls for it to be made from <110> Si wafer, 15-30 microns high, 20 micron wide fingers
<azonenberg>
But that depends on the aspect ratios i'm able to hit experimentally
<azonenberg>
If i'm lucky i could potentially get like a 10:1 aspect ratio which would permit 200um high fingers
<azonenberg>
other papers have done that
<azonenberg>
it depends really on how well my hardmask sticks to the wafer and how well it survives exposure to KOH
<azonenberg>
Right now sol-gel Ta2O5 appears to fill both of those very well
<azonenberg>
its almot impossible to remove, the "scotch tape test" doesnt damage the film whatsoever
<azonenberg>
And its etch rate in KOH according to several publications is usper low
<azonenberg>
usper*
<azonenberg>
super*
<azonenberg>
like angstroms per year
<azonenberg>
effectively zero
<azonenberg>
And now that i've almost solved the problem of patterning it, i should get very good results
<azonenberg>
44% KOH is extremely anisotropic - the etch rate of <110> is around 600 times that of <111> planes
<azonenberg>
So 100:1 aspect ratios are potentially feasible
<azonenberg>
that's on par with deep RIE
<azonenberg>
But i havent seen any literature using KOH for more than around 10:1 or 15:1
<azonenberg>
It really seems that KOH is a near-ideal etchant because it's fast, gives relatively smooth surfaces if you use it right, and is extremely selective
<azonenberg>
the only downside is metal ions which are a problem for CMOS but not MEMS
<azonenberg>
And while it's pretty fast it still will take a bit over 2.5 hours to do the 500um etch i need lol
<azonenberg>
wonders if he should have gotten thinner wafers to start with
<XgF>
I guess slow etchants are one of the reasons why foundries like big wafers?
<azonenberg>
You think that's slow? Lol
<azonenberg>
XeF2 will take overnight to go this fart
<azonenberg>
far*
<azonenberg>
But it doesnt have metal ions and is therefore CMOS compatible
<azonenberg>
But yes
<azonenberg>
It costs about the same amount to process a big wafer or a small one
<azonenberg>
but they make more money with big ones
<XgF>
On a commercial scale, I bet thats a bottleneck :p