How a Computer Found the Perfect Magnet Without Brute Force ✛ ✎ ✦
For a long time, materials scientists searched for alloys by trial and error, calculating thousands of combinations. Even self-learning programs often missed unusual compositions. The new approach relies on precise quantum calculation, pioneered by figures like John Bardeen. The key is that the proportion of each metal (iron, cobalt, nickel) can be varied smoothly, like turning a dial. At each step, the computer calculates the magnetic and electric properties of the mixture and instantly determines which way to ‘turn’ to improve the result. It’s like descending a mountain in the fog: no clear path, but the slope underfoot guides you to the valley. The most unexpected twist: in just a couple of dozen steps, the program derived the same ideal recipe, an iron-cobalt alloy, that metallurgists found experimentally in the 1930s. In a second task, requiring a specific energy pattern of electrons, the algorithm added a microscopic dash of tellurium, improving the metallic qualities. Such precision became possible by accounting for the dual nature of electrons: their quantum quirks.
The algorithm didn’t just find an alloy; it automatically derived the Slater–Pauling rule, which was considered a purely empirical discovery for almost a century.