I can't answer all these questions, but I do have some comments:

* Ti is very susceptible to oxidation and always needs to be deposited in
the same chamber, without breaking vacuum, as the subsequent metals. So the
Auger profiling you did is exactly the right thing to check on this.

* Strong aqueous bases are always good at competing away anything else that
is deposited on an SiO2 surface, particularly organic substances, glues,
waxes, silanizations, etc. I can only speculate that it can sometimes do
the same to metals. But there should be many people out there with direct
experience of metal resistance to alkaline anisotropic etchants, such as
the people using silicon-doped TMAH.

* Silicon nitride surfaces are 98% to 99% identical to silicon dioxide
surfaces, if one looks at the outer layer of atoms. This comes from
measuring the point of zero charge of Si3N4 in aqueous solutions, and is
supported by Auger measurements of nitride surfaces that have been exposed
to air. So, in general a nitride should behave just like an oxide. A
possible exception could be if one removes the outer oxide layer in high
vacuum, say by a sputter clean, just prior to deposition.

*Tried any scotch tape adhesion tests of the metal before EDP etch?

Luc Bousse
Caliper Technologies.


>I have been interested to read the ongoing comments regarding metal
>adhesion.  At present we are experiencing difficulties in this area.  Our
>metallisation consists of TiPdAu (500/2000/2000A) evaporated by e-beam onto
>an SiO2 coated substrate. We then etch the silicon in an EDP solution.
>
>Currently we are seeing poor adhesion of the metal after EDP etch although
>we have had good results in the past. We have had the pads analysed by Auger
>by milling through the pad to the interface.  There is no difference between
>well adhered and poorly adhered pads. There is also little oxygen in the Ti
>and no carbon at the interface.
>
>Specifically I would be grateful to hear comments on the following questions:
>
>Is there a known problem with this metallisation and EDP.? eg.An
>electro-chemical effect.
>
>Is it better, in general, to evaporate Ti at a faster or slower rate.?
>
>Are we likely to get a better bond with Ti to an oxide or nitride surface.?