RECIPES for a host of novel super alloys are being created in a Danish lab by applying the principles of genetic inheritance. It works so well that it has already led to a clutch of high-performance materials that would have taken years of laborious work to find by other means.
Physicist Jens N酶rskov and his colleagues at the Technical University of Denmark in Lyngby used their novel approach to find the pick of the crop among nearly 200,000 possible combinations of metals. In repeated tests, their approach consistently came up with the same final set of best alloys. Encouragingly, these included well known super-strong alloys with a high melting point. But alongside these familiar materials were several intriguing unknown ones that engineers had not thought of trying. 鈥淭his gives a catalogue of possibilities,鈥� says N酶rskov, who is arranging for the new materials to be studied by metallurgists.
To find their ideal materials, N酶rskov used a program known as a genetic algorithm. The program, which mimics the Darwinian process of natural selection, treats each group of four metals as a chromosome and each individual metal as a gene, and then strives to find the fittest permutation.
Advertisement
The program started with 20 chromosomes containing randomly chosen genes (metals) from 32 possibilities. It then began processes analogous to genetic crossover and mutation, mating different genes and randomly substituting individual metals within groups. Of the new population of 60 chromosomes, the 20 with the best properties were set aside before they too went through mating and mutation again.
The program evaluated the fitness of the chromosomes using what鈥檚 called density function theory (DFT), which predicts the properties of materials based on knowledge of how the electrons of its various kinds of atom interact. Those with the most densely packed electrons were more likely to be stronger and have higher melting points, and were deemed fitter.
The conventional way to look for new alloys is to blend together likely combinations of metals and test them, a process that is both time-consuming and hit-and-miss. In repeated tests of his new approach, N酶rskov found that after the populations were allowed to evolve for 45 generations, they ended up with the same set of fittest alloys. 鈥淵ou can鈥檛 imagine another way of doing it without doing 200,000 experiments,鈥� he says.
Though N酶rskov has only used his technique for singling out strong alloys, he says it doesn鈥檛 have to stop there. It could be used to discover materials with desirable properties such as high heat conductance.
But Colin Humphreys, a materials scientist at Cambridge University, warns that the limitations of the DFT technique mean that N酶rskov鈥檚 methods won鈥檛 work for some other properties: 鈥淭he fracture point, for example, is just too complex.鈥�