For the seed layer I do an e-beam evaporation of titanium and then gold.  After
developing the KMPR I do a quick RIE descum for 2 min with 100 W, and 10 sccm
O2 to make sure I get any latent crud out of the holes.

That second suggestion you mention seems to make the most sense.  My holes are
fairly spaced out.  Also, I am using an anode that is much larger than the area
on the wafer I am trying to plate, and I was worried that the difference in the
anode/cathode area would cause more e-field at the edges of the holes.  Is it
best to have an anode:cathode ratio that is 1:1?

Is this problem just as severe for other electroplated metals or is it more
prominent when electroplating gold?

Thanks for all the suggestions!
Parrish

Quoting Oray Orkun Cellek :

> Dear Parrish,
>
> Probably photos would be far more helpful. I had a similar experience even
> without any bubbles. Your post dimensions are important but separation of
> clear areas on the mask should also be considered.
>
> Among many causes one may be that your openings are starving.  When the mask
> design is not uniform or is not periodic, or if unintentional wide openings
> are present, large clear areas sink the electric field lines in the
> electro-plating solution.  If you have clear fields around your intended
> area, they may steal the e-field.  (That's what people call "thief mask"
> which is sometimes used intentionally to equalize e-field on the wafer).  If
> you have such large clear areas I would coat those first.
>
> Second cause would be the large separation between posts. This is opposite
> of the previous case but the result is same.  This is also logical because
> if there is too much dark area between the open areas, too much field
> accumulates at the edges of your openings.  In this case I would put dummy
> openings between the posts.
>
> Other cause may be the already low e-field or hard-seeding in the solution.
> In my experience this kind of hard seeding causes edges to rise rapidly.
> And once they rise, they keep going on because electric field lines are
> denser there.  According to my experience I would increase the pulsed
> plating peak x5 to x10 times and  wait to seed every point, and then lower
> the current density to a better value.
>
> If you still have rising edges, I would segment the post openings to smaller
> patches if the device design allows.
>
> After-plating solution would be reflowing the electroplated posts with RTP
> to make it fill the center itself (if you allow rounder shape for the posts)
>
> And lastly, what is your wetting layer ? What kind of cleaning do you do
> prior to plating ?
>
> Good luck,
>
> Oray Orkun Cellek