Flatland magnets are switched under pressure

When Xiaodong Xu of the University of Washington performed some of his first measurements on chromium triiodide two years ago, he was met with a baffling surprise: The material seemingly knew how to count.

As a van der Waals material, CrI₃ is made up of atomically thin layers that barely resist being pulled apart; it was one of the first such materials shown to be ferromagnetic in its two-dimensional form (see Physics Today, July 2017, page 16 ). When a single CrI₃ layer is cooled below its Curie temperature of 45 K, all the chromium spins align in the same direction. The bulk material, also ferromagnetic, behaves similarly, albeit with a slightly higher Curie temperature of 61 K. In a stack of two layers, however, the spins in each layer ordered ferromagnetically, but the coupling between them was antiferromagnetic: One layer was entirely spin up, and the other was entirely spin down, so the bilayer as a whole had no net magnetization.

The subtleties of reduced-dimensional systems mean that the properties of a van der Waals material aren’t necessarily retained when one or a few layers are isolated from the bulk. Even so, there was no apparent explanation for how the feeble van der Waals forces between layers could give rise to either ferromagnetic or antiferromagnetic coupling, depending on whether the layers were part of a larger structure.

A handful of theory papers have since suggested a solution. Successive layers of CrI₃ can stack in two ways, monoclinic and rhombohedral, as shown in the figure (for clarity, only the Cr atoms are shown). The bulk material is monoclinic at room temperature and becomes rhombohedral when cooled below 220 K; the rhombohedral phase then ferromagnetically orders below 61 K. Perhaps a few-layer stack doesn’t change structure upon cooling, and monoclinically stacked CrI₃ layers couple antiferromagnetically.

Now two experimental groups—one led by Xu, the other by Jie Shan and Kin Fai Mak of Cornell University—have shown that the stacking order, and thus the magnetic coupling, in few-layer CrI₃ can be controlled by applying hydrostatic pressure.

Both groups used quantum tunneling measurements and magnetic circular dichroism microscopy to measure the material’s magnetic properties, and they used polarized Raman spectroscopy to probe its stacking order. Pristine films of up to five layers ordered antiferromagnetically and had monoclinic structure. But when placed in a piston cell and subjected to pressures of 1.8–2.7 GPa, the films switched to ferromagnetic ordering and rhombohedral structure. The change persisted even when the pressure was removed.

The results suggest that it might be possible to engineer new spin textures by manipulating the stacking order. For example, when two layers of a lattice are deliberately misaligned by a small angle, they form a quasiperiodic moiré pattern. In bilayer graphene, the misalignment gives rise to unconventional superconductivity (see Physics Today, May 2018, page 15 ). Exploring the magnetic properties of twisted bilayer CrI₃ is a logical next step. (T. Li et al., Nat. Mater., 2019, doi:10.1038/s41563-019-0506-1 ; T. Song et al., Nat. Mater., 2019, doi:10.1038/s41563-019-0505-2 .)