An ASI is a lattice of nanomagnets in which long-range interaction between the components produces a geometrically frustrated system. As analogues of otherwise hard-to-study physical phenomena, ASIs have been invaluable, providing insight into areas as diverse as magnetic monopoles and thermodynamics.

Investigations are also motivated by more practical applications, however: magnonic crystals – in which information is encoded in spin waves – could underpin future generations of data storage and processing devices, for example. This potential will only be realized if the nanomagnet constituents of ASI can be manipulated with sufficient reliability and freedom.

Experimental serendipity

The discovery that an MFM can offer the necessary degree of control was made almost by accident. "What prompted our work was stumbling upon a method for injecting domain walls into nanowires when using stronger MFM tips," Gartside explains. "Using overly strong MFM tips by mistake was a good example of a happy accident in the lab leading to some nice results."

That happy accident led the researchers to a technique by which a scanning-probe tip can be made to induce a localized polarity change in a material. Gartside and colleagues found that moving an MFM tip over a nickel–iron (permalloy) nanowire produced a magnetic vortex-like structure in the underlying surface. Subsequently the team found that the vortex injection process can initiate a magnetization reversal of the whole nanowire. It is this recently developed method – which the researchers call topological defect-driven magnetic writing (TMW) – that promises such precise control over the magnetic configuration of ASIs.

Targeted approach

Until now, techniques to reconfigure ASIs have generally relied on large-scale thermal activation and annealing, and on the application of global magnetic fields. Although effective at moving the system from one gross state to another, the lack of specificity in these approaches makes the great majority of microstates inaccessible. More targeted scanning-probe methods have been used with promising results, but still need global fields to work, and may suffer from reduced functionality when adjacent nanowires interact strongly.

The researchers demonstrated the possibilities afforded by TMW by reversing the direction of selected macrospins in a simple kagome ASI "rosette". This structure, comprising 30 permalloy nanowires of length 1 µm, width 75 nm and thickness 10 nm, has 230 possible microstates, only two of which (one of each chirality) represent the system’s ground states. The likelihood of finding either of these ground states by untargeted methods (thermal activation and annealing, for example) is vanishingly small, but with the ability to repolarize nanowires at will, for Gartside and his team it was an almost trivial task. "It is testament to the power and flexibility of the writing technique that the ground state, previously eluding researchers for years, may now be written to order in an afternoon," says Gartside.

Exotic states

Another kind of state inaccessible to the thermalization technique, but now achieved using TMW, is the class that exhibits negative effective temperatures. This counterintuitive condition is possible because the ability to flip macrospins on demand allows the entropy of an ASI to be minimized even as the overall energy of the system is raised: in classical thermodynamics, adding energy to a system always increases its entropy.

The availability of exotic magnetization configurations will help make ASI even more versatile as a model system for varied physical phenomena, and will likely lead to insights not possible before the development of the TMW method. Gartside, however, is especially excited by the more tangible applications: "Particularly research into reconfigurable magnonic crystals, allowing magnetic networks such as artificial spin ice to be used as magnonic transistors, filters and other processing elements, and also research on training artificial spin ice and related networks to function as hardware neural networks. This is something that the field has wanted to explore for a long time but lacked the writing method to do so. I am also very excited to see what others might decide to do with the technique."