Dr. Isaak Baber wrote about a problem with soldering a cap to a cylinder
hermetically.

Unfortunately the text doesn't make any statements about the size and
mechanical stability of the assembly but I think, the best way is to get
rid of the flux. One possibility I could imagine to work is to have tinned
( or tin-leaded don't know how to say in English) surfaces and having the
two parts produced with mechanical tolerances to form a slight press-fit.
After assembling the two parts a pass through a reflow oven should be
sufficient to form a solder joint. This will still leave the problem of
oxidized tin. One solution would be to assemble the parts in a reducing
atmosphere or in nitrogen after having the parts held in a reducing
atmosphere for a while ( complicated ).
A weird idea I had would be to cover the leaded surfaces with gold. One has
to care about the thickness of the gold compared to the solder volume
though. I know, this does mean that the parts have to be assembled within
reasonable time after plating (growth of intermetallics to the surface
which will make soldering difficult) but I think there would be enough
time. Thanks to Phil. I know now how to spell intermetallics. Maybe he
could give a comment about the growth-rate of the intermetallics.
Or maybe, it is sufficient to simply press the two parts together since
while doing so both surfaces will be deformed cracking the tin-oxide layer.

Best regards

Guenter Grossmann