McCallum's work focuses on introducing atoms of elements into the structure that are sufficiently foreign to make the cerium atoms less profligate with their electrons. Even if that succeeds, cerium is never going to make magnets as good as Neo - but then it doesn't have to. "If you look at the difference between rare earth magnets and non-rare earth magnets, there is a tremendous gap," he says. Any magnet made of friendlier materials that sits anywhere in the gap will take some of the heat from neodymium - and above all from dysprosium.
Constantinides has a similar goal. "We don't need to replace Neo," he says. "We need other materials with a complementary price, performance and ability." His company is following two broad approaches. One is to use computational muscle to try out tweaks to existing non-rare-earth magnets such as Alnico. The other is what he describes as a greenfield approach, using complex algorithms to mix up nature's limited library of magnetic elements in different structures and analyse the results for magnetic stardom. "We've had nickel, iron and cobalt for a long time, but the question is how we can combine these elements in a clever way and come up with something really good," he says. He is by no means assured of quick success: "It takes a lot of teraflops."New Scientist (sub required)