<azonenberg_lab>
It pretty much worked - the Cu delaminated slightly due to undercut but a thinner oxide layer should fix it
<azonenberg_lab>
Pattern transferred from mask to photoresist reduced 10x, then to evaporated Cu, then to Ta2O5, then to Si
<azonenberg_lab>
Now its just a matter of getting sharper edges, better adhesion, etc
<azonenberg_lab>
Right now i dont feel comfortable calling 20 microns my achievable feature size, 30-40 is more like what i can actually do with the current process
<azonenberg_lab>
my finer features didnt quite resolve
<azonenberg>
The oxide layer was too thick ( i need to try 100nm next time) so the HF undercut the maks a little, note the copper/photoresist in the background fell off due to undercut
<azonenberg>
But the oxide is still present over most places it should be (though a little thin)
<kristianpaul>
551 looks a bit messy compared to 536
<kristianpaul>
but of course, is all Si :-)
<kristianpaul>
this is taking using metallurgical microscope, right?
<azonenberg>
Yes
<azonenberg>
And yes, i am still working on getting better results
<azonenberg>
The problem is that the Ta2O5 was too thick so the HF undercut the Cu hardmask
<azonenberg>
I need to use a thinner oxide layer and maybe a little more Cu
<azonenberg>
But overall it worked surprisingly well
<kristianpaul>
indeed, well if you take look back to previos results of course !
<azonenberg>
Lol yeah
<azonenberg>
This was the first die i actually took all the way through the process that even sort of worked
<azonenberg>
So now its just a matter of tuning it to give better results
<azonenberg>
I plan to do prep (Ta2O5 + Cu deposition) on a quarter of a <110> wafer in the next week or two
<azonenberg>
so i can try some vertical etches
<azonenberg>
wants a comb drive :)
<azonenberg>
lekernel_: How far do you think i am from a working comb drive now? I have silicon patterning mostly worked out, i can deposit dielectric (Ta2O5 is actually a high-K material), and I can deposit metal if i borrow time on the evaporator
<azonenberg>
Think end of summer to end of the calendar year is feasible?