Spintronics

Electrons possess both charge and spin, giving rise to electrical and magnetic properties, respectively. Spintronics exploits these two aspects within the frameworks of electronic and magnetic engineering. Building on this foundation, our laboratory focuses on the following main research directions:

Exploration of energy-efficient nanoscale spintronic devices
Development of spintronic systems for next-generation memory and computing

Nano-Spin Devices

Spintronic Materials

Spintronic materials are engineered to control and utilize the spin of electrons for generating, transporting, and detecting spin currents in functional devices. A key class is spin-source materials, such as heavy metals and topological materials, which efficiently generate spin currents via mechanisms like the spin Hall effect. In parallel, novel magnetic materials are being explored to enhance performance and enable new functionalities, including improved switching efficiency, stability, and scalability.

These material platforms provide the foundation for next-generation spintronic devices, while also opening opportunities to study unconventional magnetic phenomena—for example, noncollinear antiferromagnets have recently attracted attention for their unique spin-dependent transport properties.

In this project, students will gain hands-on experience in crystal growth and thin-film deposition, as well as in comprehensive characterization of material properties relevant to spintronic applications.

Magnetic Tunnel Junctions

Magnetic tunnel junctions (MTJs) are the fundamental building blocks of spintronic memory devices such as MRAM. An MTJ consists of two ferromagnetic layers separated by an ultrathin insulating barrier, where the relative alignment of the magnetic layers (parallel or antiparallel) determines the electrical resistance through the tunneling magnetoresistance (TMR) effect.

MTJs enable non-volatile data storage with fast switching and high endurance, making them a key component in advanced memory technologies.

In this project, students will gain practical experience in the fabrication and processing of MTJs, including thin-film deposition, patterning, and device integration for MRAM unit cells. In addition, they will explore how variations in MTJ structure influence device properties, and apply this understanding to optimize performance for device applications.

Related Journals

J. -Y. Yoon et al., arXiv:2604.14446 (2026).
T. Kinoshita, J.-Y. Yoon* et al., Appl. Phys. Lett. 127, 102408(1)-(6) (2025).
J.-Y. Yoon et al., AIP Advances 11, 065318 (2021).
J. Yoon et al., Appl. Phys. Express 13, 013001 (2020).

Spintronic Systems

Advanced Memory Systems

Advanced memory systems based on spintronics utilize the electron’s spin to achieve fast, non-volatile, and energy-efficient data storage.

MRAM, based on magnetic tunnel junctions, is a leading technology and can be implemented in crossbar arrays for high-density integration and in-memory computing. In addition, antiferromagnetic memory offers strong robustness to external fields and ultrafast switching, making it attractive for high-speed applications.

Together, these approaches provide a promising platform for next-generation memory and computing systems.

Probabilistic Computing

Probabilistic computing in spintronics exploits the intrinsic stochastic behavior of nanoscale magnetic devices to perform computation beyond deterministic logic. Unlike conventional electronics, where outputs are fixed, spintronic devices can exhibit random switching due to thermal fluctuations, and this randomness is used as a computational resource.

A key element is the probabilistic bit (p-bit), which fluctuates between 0 and 1 with a tunable probability. In spintronic systems, p-bits are typically realized using low-energy barrier nanomagnets, whose switching behavior can be controlled by electrical inputs.

By coupling many p-bits, probabilistic circuits can efficiently solve problems such as optimization, sampling, and inference—tasks important for machine learning and complex decision-making. This approach offers advantages such as high energy efficiency, inherent parallelism, and direct hardware implementation of probabilistic algorithms.

Related Journals

J.-Y. Yoon et al., Nature Materials 22, 1106-1113 (2023).
Y. Takeuchi et al., Science 389, 830-834 (2025).
J.-Y. Yoon et al., Nature Communications 16, 1171 (2025).
W. A. Borders et al., Nature 573, 390-393 (2019).

Page updated

Report abuse