Designing things like LEDs and transistors has, for a long time, been an arduous process of trial and error, but that could be changing soon, thanks in part to a technique developed by physicists at SUNY-Buffalo.
In a paper that’s just been accepted for publication by Physical Review E, the materials scientists describe a “harmonically mapped averaging” method of simulation that allows the thermal properties of a material to be accounted for, overcoming one of the major hurdles to efficient, predictive simulations of the behavior of new compounds.
When trying to create a structure with specific properties, like the P-N junction in an LED that gives off a certain wavelength of light, scientists try different mixtures of elements. While the search for the right combinations and concentrations is guided by theory, it’s often difficult to predict how a material will behave in real life. This is due to a number of factors, one of which is thermal energy. In a simulation, atoms in a solid generally stay where they’re put, effectively acting like they’re at absolute zero. In reality, the nuclei of those atoms are constantly jangling around, exhibiting brownian motion, which can affect the way they bond and interact.
In an ideal world, we could discover materials with new or unusual properties by simulating them at the atomic level. An engineer trying to create an infrared LED, for example, could tell a program what energy bandgap he’s looking for, and find a list of candidate compounds that should suit his needs. At present, this isn’t possible, but the methods outlined in this paper should bring us a big step closer to that future.