Spin waves—often called magnons—can carry information in ways that promise lower energy use than conventional electronics. By tailoring how these waves propagate through patterned magnetic materials, researchers are building magnonic circuits that may enable compact logic, memory, and even physically inspired neural computing.
One of the most versatile tools in this effort is the magnonic crystal: a magnetic material patterned with periodic structure. Like semiconductors shape electron behavior with band structures, magnonic crystals create magnonic band structures, including frequency ranges known as magnonic band gaps (MBGs) where spin waves cannot travel.
Yet turning design intent into working structures is difficult. The shape and geometry of each unit cell determine how bands form, and the relationship between lattice features and magnon dispersion can become especially complex for higher-order bands. Prior work often concentrated on lower-order bands, where outcomes are easier to predict, leaving a gap in strategies for engineering the widest “complete” band gaps.
A team led by Professor Masato Kotsugi and second-year doctoral student Ryunosuke Nagaoka has now demonstrated an inverse-design framework aimed at finding two-dimensional magnonic crystal topologies with unusually large complete MBGs. Their work, reported in Small Structures (published July 28, 2026), introduces an optimization pipeline that searches far beyond conventional geometry rules.
The method combines frequency-domain micromagnetic simulations with topology optimization. Specifically, it evaluates candidate band structures using the frequency-domain Landau–Lifshitz–Gilbert (FD-LLG) equation, which estimates magnonic dispersions more efficiently than standard time-domain simulations—an advantage when screening many designs.
To explore the enormous design space, the researchers encode the material distribution inside a unit cell as a binary vector and run a genetic algorithm (GA) for global optimization. Each GA cycle proposes a new lattice, runs FD-LLG to score the resulting MBGs, and then evolves the unit-cell representation toward better candidates, repeating this loop until convergence.
Optimization revealed a non-intuitive topology that substantially outperforms conventional designs, with the largest band gap occurring between the fourth and fifth bands. To interpret why these structures work, the team applied an explainable machine-learning-based analysis to visualize the design landscape, finding that it becomes increasingly non-convex at higher-order bands—suggesting multiple viable solutions.
Beyond the immediate performance gains, the approach offers a general recipe for engineering magnonic crystals in experimentally accessible materials and dimensions. If higher-order-band design rules mature, magnonic platforms could support multi-frequency signal processing, faster spin-wave communication, and energy-efficient spintronic hardware—potentially relevant to future data-center architectures.
Subject of Research:
Not provided.
Article Title:
Inverse design of two-dimensional magnonic crystals via topology optimization with frequency-domain micromagnetics
News Publication Date:
28-Jul-2026
Web References:
https://doi.org/10.1002/sstr.70526
References:
DOI: 10.1002/sstr.70526
Image Credits:
Credit: Professor Masato Kotsugi from Tokyo University of Science, Japan
Keywords:
Magnonic crystals; spin waves; magnons; band gaps; topology optimization; frequency-domain micromagnetics; Landau–Lifshitz–Gilbert equation; genetic algorithm; machine learning; spintronics
