Optimising the production of a spintronic system with multiple sub-nanometre thick layers

09 Sep 2026 - Rosie de Laune

A group of researchers from the University of Leeds, Imperial College London, ISIS and Diamond Light Source have optimised the build of a complex multilayered system with potential for spintronics applications.

Schematic and 3D model of a magnetic heterostructure stack consisting of a Cap, six repeating Ru (0.5 nm) / CoB (0.6 nm) / Pt (0.8 nm) trilayers, a Buffer layer, and a Bi₂Se₃ (20 nm) base, with a color map representing out-of-plane magnetization ($m_z$) from -1 (red) to +1 (blue).

Schematic representing standard sample structure grown on Bi2Se3 with illustrated skyrmion spin texture in the CoB layers

Magnetic multilayers (MMLs) are made of ultrathin layers of magnetic and heavy metals. They are known to host multitudes of exotic magnetic phenomena, that are technologically relevant phenomena such as chiral spin textures (skyrmions) and topological insulators (TIs). TIs have long been of interest in next-generation spintronic devices because of their highly efficient charge-to-spin conversion.

This study combines both in an intricate, multilayered system. Each offers unique advantages for spintronics applications, and being able to combine the two provides an opportunity to make the most of the strengths of both materials, with the aim of reducing electronic device current densities and enhancing energy efficiency.

Ben Brereton, who is a Facility Development student joint between ISIS, Diamond Light Source and the University of Leeds, is first author on the paper, published in Physical Review Materials, which aimed to optimise the production of such a combined system. The group combined 20 nm epitaxial layers of Bi₂Se₃, a well-known TI, and a six-layer [Pt(0.8 nm)/CoB(0.6 nm)/Ru(0.5 nm)] MML with a buffer layer in between.

To optimise the process, they looked at two different Bi₂Se₃ samples made with narrow and wide surface terraces. Wider terraces are preferred, as it reduced the number of terrace edges where magnetic domains can get trapped. They also investigated the effect of the buffer layer between the TI and the MML, using tantalum (Ta) and molybdenum (Mo) at different thicknesses.

Using polarised neutron reflectometry on PolRef, they were able to see that the desired perpendicular magnetic anisotropy was consistent across the six layers of the MML. When considering the different terrace structures, they saw these were carried through into the MML layers. As these terrace structures are so important for the domain ground state, the morphology of the TI is critical to the overall device performance.

However, they discovered that so is the buffer. If none is present, then they saw the MML break down, because of intermixing between the layers. They also saw that domain ordering was better defined when using a 2.0 nm Ta buffer when compared to the 1.0 nm thick Mo buffer. However, this had a smaller influence than the terrace structure of the TI.

With the knowledge of the effect of both terrace size and buffer composition and thickness, they were able to optimise the system to achieve a labyrinthine domain ground state at zero field, which is promising for hosting skyrmion textures. The continuous multilayer structure that they produced is expected to reduce the trapping of domain walls and topological spin textures at the terrace edges, essential for any future device applications.