Studies of Artificial Spin Ice at Brookhaven and Lawrence Berkeley National Laboratories

Artificial spin ices are arrays of patterned nanomagnets. They make use of nanotechnology to reproduce many of the features found in the 'naturally-occurring' spin ices--pyrochlore crystals, in which rare earth moments are arranged meeting in groups of four at tetrahedral vertices so as to enforce geometric frustration--in a two-dimensional analog. In these artificial systems, all parameters can be specified by design and the exact configuration of the moments can be inspected using high resolution magnetic microscopy. As a result they offer exciting possibilities in studying statistical mechanical models in and out of thermal equilibrium, visualising phenomena such as fractionalised monopole excitations at room temperature in a way that is impossible in the pyrochlore systems, and exploring novel applications, such as using the 'magnetricity' of these monopoles to perform neuromorphic computing.

Fabricating and studying these systems requires state-of-the-art facilities. Here we seek travel support to allow us to access three unique facilities at two US National Laboratories: the Center for Functional Nanomaterials (CFN) and the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (BNL) and the Advanced Light Source (ALS) at Lawrence Berkeley National Laboratory (LBNL), at all of which we have won facility time to study artificial spin ices through competitive peer review. We will fabricate samples at the extreme nanoscale at the CFN, and study them in reciprocal (NSLS) and real (ALS) space at the two synchrotron facilities.

These artificial magnetic metamaterials are controlled examples of frustration, a phenomenon that is found throughout condensed matter physics and beyond. Aside from its traditional habitat of magnetism, it can describe models of protein folding, for instance, and even social phenomena such as financial networks. Moreover, these artificial systems are a workbench for exploring statistical mechanical models in a unique way where all the parameters can be tuned at will during the lithography, and the exact microstate of the statistical mechanical system can be inspected using advanced microscopy methods. Since these are systems that can be driven controllably between thermal and athermal states, they can be used to explore the physics of 'Matter far from equilibrium', an area highlighted by EPSRC under a Physical Sciences signpost and as a Grand Challenge topic.

We can expect impact in a number of areas. The most obvious are in industries with an interest in nanoscale magnetic structures (such as the data storage, microelectronics, and sensor industries). The data storage industry in particular is readying itself to switch from conventional continuous media, where the written bit comprises a collection of small grains, to bit-patterned media where exactly one bit is stored in each of many magnetic nanostructures that form an ordered array. This drastically reduces transition noise between bits, as the edges are now very well defined. However there are obvious fabrication challenges. There is also the fact that most ordered arrays of magnetic elements have an antiferromagnetic ground state enforced by magnetostatic interactions between neighbouring bits. These interactions will become stronger as bits are more densely packed, making any written data sequence unstable against this lower entropy ground state. Understanding how to frustrate these interactions will permit higher data densities.

These systems also resemble magnetic quantum cellular automata that can be used to carry out very low power logic functions. Indeed, elementary excitation out of the ground state can be described as fractionalised magnetic monopoles (of both signs of magnetic charge) that are a natural candidate for the representation of digital data. The naturally occurring spin-ices hence act as magnetolytes and the motion of these monopole excitations under field has been dubbed 'magnetricity'. Whilst not being useful for transmitting power, we can imagine using the monopole excitations that we will study in this project to convey and process information using artificial magnetricity.

This monopole computing concept is still essentially computing under a von Neumann architecture. Much further in the future we can imagine exploiting the inherent frustration in these systems to develop neuromorphic computing of the Sherrington-Kirkpatrick type in a designer magnetic metamaterial.