Faster magnetic tunnel junctions for random number generators

Random number generators (RNGs) are important for a wide range of applications, from encryption to real-world simulations. Generating truly random data sets can be computationally expensive and time-consuming, and thus magnetic tunnel junctions (MTJs) — which convert magnetic information into electrical information — have emerged as a solution for tunable RNGs.

Hamid et al. studied two newer MTJs — ferrimagnetic (Fi-MTJs) and noncollinear antiferromagnetic (NC-AFMTJs) — compared with the widely-used ferromagnetic tunnel junctions (F-MTJs) to determine their efficacy.

While all these MTJs provide the tunability RNGs need for many applications, Fi-MTJs and NC-AFMTJs have lower net magnetizations than F-MTJs, which makes them faster.

“This paper is trying to analyze how can ferrimagnets — and these very new non-collinear antiferromagnets — provide the speed up you need,” said author Jean Anne C. Incorvia. “You really want to go to picoseconds in order to get the value proposition.”

By modeling these three types of MTJs, the researchers discovered that NC-AFMTJs are about 10 times faster than the Fi-MTJs, which in turn are about 10 times faster than F-MTJs. While the non-collinear antiferromagnetic materials known so far transition into collinear antiferromagnetic states at around 86 C, ferrimagnetic materials can withstand temperatures of up to 185 C or higher, showing a tradeoff of temperature and speed can determine which MTJ type is best.

Author Shafin Bin Hamid hopes this research will serve as an “analytically driven value proposition for continuing to do research on these new and interesting types of magnetic materials for RNG applications.”

“This is all analytical modeling,” he said. “We’re trying to distill down how they can behave, [but] I think what we can show is that they can still be random.”

Source: “Ferrimagnetic and non-collinear antiferromagnetic tunnel junctions for random number generation and probabilistic computing: A comparative study,” by Shafin Bin Hamid, Alexander N. Chin, and Jean Anne C. Incorvia, Journal of Applied Physics (2026). The article can be accessed at https://doi.org/10.1063/5.0346471 .