This may be useful in actually creating something I theorized about a few months ago. I call it a chemomagnet (as opposed to electromagnet).
The basic premise is that individual magnetic domains are free to float and are unorganized until a chemical signal locks them into place in some sort of structure. The energy required to align and bind them would be chemical in nature, and once bound, another chemical signal could be sent to unbind them.
The magnet would be controllable, like an electromagnet, but would not require any electricity at all to operate.