Robert,

Functionally, anodic bonding at the fairly extreme vacuum you are
referring to is not functionally different from that at more moderate
vacuums (10e-1 torr, for example).  The biggest question is how to
arrange a system that will allow you to:

A) give you the capability to allow a pathway (i.e. a space) between
your substrates to allow removal of atmosphere, while
B) Giving you the capability of removing the spacing once the vacuum is
established, while
C) Still allowing the high level of vacuum to be established

Once the vacuum is established, the same process flow will occur:

1) Heat substrates to bonding temperature (~350C is a good starting
point)
2) Apply voltage (if you are talking die, you probably don't need too
much) - of course, you will need some way to introduce the probe that will not
violate vacuum

This is actually a pretty standard process from EVG's point of view -
our bonders have 7.5e-6 torr as an available option (pretty typical
option, to boot), and have several customers performing high-vacuum
anodic bonding.

Best Regards,
Chad Brubaker