Researchers at the University of Waterloo recently demonstrated fully electrical, field‑free control of perpendicular magnetization using spin‑orbit torque (SOT) in a low‑symmetry 2D magnet/topological‑insulator heterostructure, paving the way for scalable, energy‑efficient spintronic memory and logic devices.
Stacking the three-fold symmetry of BiSbTe on top of the two-fold symmetry of intercalated-CrTe, the interface only permits a unidirectional symmetry which produces an extremely strong out-of-plane spin torque and can deterministically switch a very hard, perpendicular magnet with ease. Image credit: University of Waterloo
Modern MRAM and related spintronic memories need dense, robust perpendicular magnetic anisotropy (PMA) bits that can be switched deterministically with low energy consumption, but conventional SOT easily switches only in‑plane moments and typically requires an external bias field to tilt perpendicular spins “up” or “down”. In perpendicular configurations, bits point out of the film plane, which boosts storage density but makes the energy‑efficient, fully electrical control of their state difficult. Standard heavy‑metal/ferromagnet stacks already break out‑of‑plane symmetry and can support in‑plane switching, yet deterministic out‑of‑plane reversal demands breaking additional in‑plane symmetries - usually via an applied magnetic field, which adds circuit complexity, power overhead, and risks cross‑talk between neighboring bits.
To address this issue, the researchers engineered a heterostructure combining a self‑intercalated 2D magnet, Cr3Te4, with a topological insulator, (Bi0.75Sb0.25)2Te3, grown by molecular beam epitaxy (MBE) on wafer scale. The topological insulator provides spin‑momentum–locked surface states that convert charge current into a strong spin accumulation, while the Cr3Te4 layer exhibits perpendicular magnetic anisotropy and an ordered 2 × 1 self‑intercalation pattern. This ordered intercalation breaks all in‑plane symmetries at the interface and yields an effective unidirectional m (Cs) symmetry, which is exactly what is needed to make the SOT deterministic for out‑of‑plane switching without any external field. Unlike exfoliated 2D single crystals, their wafer‑scale films naturally nucleate three equivalent 2 × 1 sublattices, so the net SOT switching exhibits a characteristic three‑fold angular dependence - an experimental fingerprint of the underlying interfacial symmetry.
In terms of performance, the heterostructure supports extremely strong, field‑free SOT switching of Cr3Te4 with perpendicular coercivity of about 1.3 T, showing that even very “hard” PMA magnets can be flipped purely electrically. The giant SOT originates from the high charge‑to‑spin conversion efficiency of the topological surface states in (Bi0.75Sb0.25)2Te3, which generate a large spin current from an applied in‑plane charge current. Because the interfacial structure has deliberately engineered low symmetry due to the 2 × 1 self‑intercalation, this spin current acquires an out‑of‑plane component that acts as a unidirectional torque on the magnetization, selecting the “up” or “down” state without needing a bias magnetic field. The same interface design allows the team to tune both the magnitude and direction of the SOT, while the presence of multiple sublattice domains emphasizes the importance of thin‑film growth control and domain engineering for practical SOT devices. As Professor Guo‑Xing Miao put it, “By engineering the interfacial symmetries, we have achieved a giant, perpendicular spin-orbit torque that can switch even very hard magnets without actual applied field,” underscoring the central role of interfacial symmetry engineering.
This work provides a route to bias‑field‑free, fully electrical switching of PMA bits with coercivities exceeding 1 T, using simple heterostructures compatible with wafer‑scale growth. It brings integrated, SOT‑driven MRAM and logic devices closer to reality and suggests that similar low‑symmetry interface strategies could be extended to 2D quantum materials and future targets such as altermagnets for ultrafast, ultra‑dense memory technologies.