Note that for high temp operation in device life, above 600 C, Ta might
be a better adhesion layer, or should be used as a barrier.  Ti alone
can oxidize further as it and/or the oxide diffuses up through the Pt
layer, because oxygen will diffuse through the Pt.

There is one particularly good paper on the Ta substitution, but I don't
have it in front of me.  One that mentions the Ti diffusion issue is:
Fox GR.  Trolier-McKinstry S.  Krupanidhi SB.  Casas LM.; "Pt/Ti/SiO/sub
2//Si substrates"; Journal of Materials Research, vol.10, no.6, June
1995, pp.1508-15.
Abstract
  Pt/Ti/SiO/sub 2//Si structures have been studied to
  investigate the structural, chemical, and microstructural changes that
occur during annealing. Grain growth of the as-deposited Pt columns was
observed after annealing at 650 degrees C, and extensive changes in the
Pt microstructure were apparent following a 750 degrees C anneal for 20
min. In addition, two types of defects were identified on the surfaces
of annealed substrates. Defect formation was retarded when the surface
  was covered with a ferroelectric film. Concurrent with the
annealing-induced Pt microstructure changes, Ti from the adhesion layer
between the Pt and the SiO/sub 2/ migrated into the Pt layer and
oxidized. It was shown with spectroscopic ellipsometry and Auger
electron spectroscopy that for long annealing times, the titanium oxide
layer can reach the Pt surface. Consequently, at the processing
temperatures utilized in preparing many ferroelectric thin films, the
substrate is not completely inert or immobile. The changes associated
with Ti migration could be especially problematic in techniques that
require the substrate to be heated prior to film deposition. (20
  References).

Charles Lakeman wrote:

>I'd add that Pt is capable of withstanding high post-deposition
>temperatures.  Pt/Ti/SiO2/Si stacks were developed for use as a bottom
>electrode for ferroelectric film capacitors that often require a post
>deposition anneal at T>600C.  There is a wealth of information on Pt/Ti
>interdiffusion, adhesion, stress development, microstructure, etc. in the
>ferroelectrics literature.  For a start check out the MRS proceedings
>Ferroelectric Films (I think there are now something like 13 volumes over
>the past 15+ years).
>
>Charles D.E. Lakeman, Ph.D
>Senior Scientist
>TPL Inc.
>3921 Academy Parkway North, NE
>Albuquerque, NM 87123
>505/342-4427 (Tel)
>505/343-1797 (FAX)
>
>
>-----Original Message-----
>From: [email protected]
>[mailto:[email protected]] On Behalf Of
>David Nemeth
>Sent: Tuesday, December 09, 2003 9:14 AM
>To: General MEMS discussion
>Subject: RE: [mems-talk] Why Choose Platinum
>
>It depends on the purpose.
>
>A thin layer of platinum is an excellent diffusion barrier (Ti is not so
>good at this) to prevent interdiffusion of metal into Silicon and vise
>versa.  For example, if you want a conductive layer, you might use thin Ti,
>thin Pt, and thick gold.  The platinum prevents the gold from diffusing into
>the silicon, and prevents the silicon from diffusing into the gold.
>
>If you are making a temperature sensor, platinum has a nice, linear
>resistance versus temperature down to very low temperatures (about 10
>Kelvin).
>
>David Nemeth
>Senior Engineer
>Sophia Wireless, Inc.
>14225-C Sullyfield Circle
>Chantilly, VA