The only way to get this kind of data through most XML parsers is to remove or replace the offending character first.

XML has strict rules about what characters are allowed (independand of character encoding, it seems), and anything that doesn't conform is not considered to be XML and parsers should fail when they encounter invalid XML. The theory is that it's better to croak on parsing than to let through a potentionally corrupt file. Besides that, the parser has no way of knowing what the intended character should be anyway.

If you want to push binary data through, maybe using a &#number; code will help. I haven't tried it.