Tokyo University of Science Reveals How Maze-Like Magnetic Patterns Form and Evolve

Breakthrough in Understanding Maze-Like Magnetic Patterns

Researchers at Tokyo University of Science have made a significant advance in materials science by developing a new computational model that explains the formation and evolution of maze-like magnetic patterns in soft magnetic materials. These intricate structures, known as maze magnetic domains, play a crucial role in the performance of devices like electric motors and transformers. The study, published in Scientific Reports, addresses long-standing questions about how these patterns emerge during magnetization reversal processes, particularly under varying temperatures.

Soft magnets, characterized by their low coercivity and ability to easily magnetize and demagnetize, are essential in applications requiring efficient energy conversion. However, a key challenge is iron loss, or hysteresis loss, which occurs when magnetic fields repeatedly reverse direction, generating heat and reducing efficiency. In electric vehicles and renewable energy systems, minimizing this loss is vital for sustainability.

The Mystery of Maze Magnetic Domains

Maze magnetic domains appear as complex, zig-zag patterns in thin films of materials like rare-earth iron garnets, or RIG. These perpendicularly magnetized films exhibit branching and curving walls that intensify with temperature increases from room conditions up to 80 degrees Celsius. Traditional theories, such as the Ginzburg-Landau framework, assume uniform magnetization and overlook thermal fluctuations, while micromagnetic simulations like the Landau-Lifshitz-Gilbert equations neglect entropy's role.

Experimental observations using polar Kerr microscopy reveal these domains shifting dramatically near the coercivity point—the field strength where magnetization reverses. Yet, pinpointing the causal mechanisms behind nucleation, propagation, and pattern complexity has proven elusive due to the interplay of exchange interactions, demagnetizing fields, and thermal entropy.

Innovative eX-GL Model from Tokyo University of Science

Led by Professor Masato Kotsugi and Dr. Ken Masuzawa from the Department of Materials Science and Technology, the team introduced the entropy-feature-eXtended Ginzburg-Landau, or eX-GL, model. This physics-informed approach integrates configurational entropy into the Helmholtz free energy equation: F_total = E_demag + E_exch - T S, where E_demag is demagnetization energy, E_exch is exchange energy, T is temperature, and S is entropy.

By treating the magnetized state as an Ising spin system, they approximated entropy from pixel-wise magnetization distributions in binarized Kerr images. Pseudo-energies were computed considering neighbor interactions and dipole sums, enabling a realistic energy landscape.

Data-Driven Pipeline: Persistent Homology and Beyond

The model's strength lies in its data-driven pipeline. Persistent homology, a topological data analysis tool, extracts generators from persistence diagrams of domain images. These capture multi-scale features: near-diagonal points indicate stripes, off-diagonal ones zig-zags, and corner clusters texture variations.

Persistence images serve as feature vectors, reduced via principal component analysis to PC1 and PC2—the primary order parameters. PC1 correlates quadratically with net magnetization M, linking microstructure to macro properties. Gradients in the PC1 energy landscape reveal reversal dynamics.

Key Findings: Four Energy Barriers Unveiled

The eX-GL landscape identifies four critical energy barriers during reversal:

• Barrier I: Nucleation requires overcoming all energy components to break uniform alignment.
• Barrier II: Maze formation demands exchange and entropy input for domain widening.
• Barrier III/IV: Demagnetization relief transfers energy to exchange and entropy, proliferating walls and boosting complexity.

Temperature flattens the landscape near coercivity, easing transitions. Domain wall lengthening drives entropy rise, coupling structure to loss mechanisms. At higher temperatures, fewer but more branched generators form, explaining observed pattern evolution. Read the full study in Scientific Reports.

Temperature-Dependent Behavior and Entropy's Role

Experiments on 475-micrometer-thick RIG films from 0-80°C show domains simplifying at low temperatures (stripe-like) and complexifying higher up (branching mazes). The model quantifies this: entropy S surges with disorder, stabilizing reversed states despite higher exchange costs from elongated walls.

PC1 rises linearly with temperature at coercivity, expanding the low-energy flat region for easier reversal. This thermal facilitation aligns with partial demagnetization in operating motors, where heat exacerbates losses.

Implications for Energy-Efficient Technologies

Hysteresis loss constitutes 31% of motor iron losses, critical for electric vehicles where efficiency gains translate to longer range. By revealing entropy-domain wall coupling, the eX-GL model guides defect engineering to suppress unnecessary branching, minimizing switching energy.

Applications extend to transformers, inductors, and sensors. In Japan, with its push for carbon neutrality by 2050, such optimizations support high-efficiency motors in EVs and renewables. Professor Kotsugi notes: "This framework deciphers hidden mechanisms, paving the way for low-loss materials."

Tokyo University of Science's Materials Research Legacy

Tokyo University of Science (TUS), founded in 1881, excels in applied sciences. The Department of Materials Science and Technology focuses on nanomaterials, magnetism, and informatics for green energy. Recent works include AI-accelerated single-molecule magnets and causal magnetization analysis.

Funding from JSPS KAKENHI and JST-CREST underscores Japan's investment in materials informatics. Collaborators from Tsukuba, Okayama, and Kyoto highlight interdisciplinary strength.

Broader Context: Soft Magnets in Japan's Tech Landscape

Japan leads in soft magnetic materials, producing non-oriented electrical steel for motors. Companies like JFE Steel and Nippon Steel innovate alloys reducing losses by 20-30%. TUS research complements this, using topological tools for unprecedented causal insights.

Related studies explore magnetoelastic effects shrinking hysteresis loops and AI identifying loss origins in steel. Globally, EV demand (projected 40 million units by 2030) amplifies need for such advances.

Future Directions and Model Generalizability

The eX-GL framework extends beyond Ising models to XY/Heisenberg systems, applicable to skyrmions, ferroelectrics, and crystal growth. Future work: integrate with atomistic simulations for defect design; apply to polycrystalline steels.

TUS plans experimental validation via synchrotron imaging. This could cut motor losses 10-20%, aiding Japan's 46% emissions cut by 2030.

Photo by Hirzul Maulana on Unsplash

Stakeholder Perspectives and Real-World Impact

Industry experts praise the causal visualization: "Automated barrier detection revolutionizes materials optimization," per a motor engineer. Academics note its explainable AI edge over black-box ML.

For students, TUS offers programs in materials informatics, fostering Japan's next innovators. Explore opportunities in Japanese higher ed for cutting-edge research.