What’s inside a magnet?
The tendency of electron spins to align with their neighbors is the defining feature of ferromagnetism. But not all stable magnetic configurations have global alignment. Magnetization on the submicron scale can organize into domains, vortices, skyrmions, and other inhomogeneous configurations. (See, for example, the article by Chia-Ling Chien, Frank Zhu, and Jimmy Zhu, Physics Today, June 2007, page 40 .)
Those structures are usually studied in thin magnetic films, where the magnetization is effectively two dimensional. Magnetization fields of 3D objects can be even more complex, with plane-like domain walls, line-like vortex cores, and zero-dimensional singularities called Bloch points. At the core of a 2D skyrmion, the magnetization can relieve its tension with the surrounding spins by popping out of the plane. At a Bloch point, the magnetization has no choice but to go to zero.
At least, that’s what theory prescribes; the structure of Bloch points has yet to be directly observed. Nondestructive imaging of bulk magnetic materials has heretofore been limited to neutron diffraction, whose resolution, at tens of microns, is orders of magnitude too poor to see most predicted structures of interest.
Now Claire Donnelly , Laura Heyderman (both at the Paul Scherrer Institute and ETH Zürich in Switzerland), and their colleagues have developed an x-ray tomographic technique for getting a much finer view of 3D magnetization structures. Using hard x rays from the Swiss Light Source, they collected diffraction patterns of a gadolinium–cobalt micropillar at more than a thousand different orientations, then computationally reconstructed the internal magnetization field. A cross section of their results is shown in the figure. They still can’t see the Bloch points themselves—which, with a predicted radius of 5 nm, are a factor of 20 too small to be directly resolved—but they’ve obtained the first images of the singularities’ effect on the surrounding magnetization. And as hard x-ray sources become brighter, the method’s resolution will improve. (C. Donnelly et al., Nature 547, 328, 2017 .)