wrote:

it seems obvious that it should be an inflatable sphere.

Have you looked into NASA's Transhab?

A quick overview of Transhab:
http://www.astronautix.com/craft/traodule.htm

This link details its materials selections and structu
http://faculty.erau.edu/ericksol/pro.../transhab.html

Maker of the Transhab test article:
http://www.ilcdover.com/products/aer...e/habitats.htm

I'm considering a sphere 100 meters across, with an internal shell 3
meters inside that. The inner and outer shells will eventually be
augmented with many layers of fabric and epoxy, but initially are
essentially just aluminized mylar and very fragile.

You can probably start more robustly than that.

If you don't mind developing a big rocket (like Boeing's 7-core Delta
IV concept), then you can launch a sizable, robust sphere immediately.

A 100m sphere pressurized to 4.9psi with oxygen and a dash of nitrogen
would generate 61300psi of stress in a 2mm (1/12") shell. That's well
within the strength capability of a plastic like Kevlar (or the more
modern Zylon). The shell would be about 80-90 metric tons, depending on
the particular shell material selected.

Then, of course, you'd want to look into insulation, debris shields,
additional shells, etc. Scaling up from the 8.2m Transhab cylinder to a
100m sphere suggests about 500-600 metric tons of internal material
would be needed.

Interestingly, the resulting shell (all ~700 tons) is just about as
heavy as the air inside. A 100m sphere of air at sea level pressure
(80/20 nitrogen/oxygen) is 673 metric tons at 25C, if I did my math
correctly.

Actually, that's a thought...the 4.9psi (1/3-bar) initial oxygen
atmosphere I suggested would be 250 metric tons.

This level of bulk material delivery would really encourage the
development of a big launcher.

I know plastic at exremely low temperatures is extremely brittle.

Only if you use the wrong plastic. Some plastics behave very well at
cryogenic temperatures. See the Transhab development notes I linked in
above. Kevlar and polyethylene remain fairly tough and ductile at
cryogenic temperatures.

Mike Miller, Materials Engineer