In my experience modifying the EDP recipe to make it not to attack Aluminum is
very difficult. That also, chip industry use Si-Al alloy for metallization, and
this makes the resustls totally inconsistent between runs, batches and
companies.

It is lot easier to use an electroless technique to convert all exposed
aluminum to something else that can stand EDP for a while. That will certainly
save alot of time tinkering around cooking the right EDP bath.

You can use one other simple approach. Don't open the Aluminum pad area at all.
What I mean is do your design such that the pad areas are not opened in your
layout design. This way the chips from CMOS foundary will come with a thin
layer of passivation (glass) on top of the pads.

Do your post processing in EDP. This will eat away a small amount of
passivation layer all over the chip (unless it is pure nitride). Due to the
high residual stress of nitride, I was told by many CMOS foundaries that even
if they use nitride for passivation it will be just a very very thin nitride on
top and all the rest of passivation will be SiO2, or doped SiO2.

After post processing, use a ultrasonic wedge bonder with >2 times the usual
power and you can break through the passivation layer and bond to the aluminum
layer below. This technique has worked everytime I tried in the past 9 years.
We went into electroless plating because we needed to build some ethanol sensor
and that required some exposed metallization area on the suspended platform.
Obviously we cannot use the bonder to break open the passivation on a suspended
portion of a microstructure :-)

Incidentally, isn't ascorbic acid Vitamin "A"?

Sincerely

Ash