Nanomagnetic logic, in which the outcome of a computation is embedded into the energy hierarchy of magnetostatically coupled bistable nanomagnets, offers an attractive pathway to implement low-power in-memory computation. The computational operation in such artificial spin systems is driven by thermally activated reversals of magnetic moments, thus requiring little energy, and its result is retained afterwards due to the non-volatility of the nanomagnets, thus avoiding separate energy costs associated with the transport and storage of information that occur in the commonly-used von-Neumann computation architecture.
Spontaneous fluctuations and reversals of the individual nanomagnets allows an interacting system to relax to a state of lower energy, which can encode the result of a computational operation. Thermal energy increases the probability of such transitions, however, commonly used global heating schemes (e.g. via thermal contact to the sample substrate) have the disadvantage of slow operation speeds as well as lack spatial selectivity and are thus inherently wasteful. These disadvantages can be removed by employing local plasmon-assisted photo-heating in hybrid nanomagnetic-plasmonic metamaterials. These combine nanomagnetic elements with a thermoplasmonic heater that can be excited by laser light, leading to highly localised, fast, and sub-lattice specific heating up to several hundred Kelvin to enable magnetic switching.
Based on this approach, i.e. combining arrays of nanomagnets with local plasmonic heating, the LICONAMCO project aimed to develop optically reconfigurable computation schemes towards low-power ultra-fast computing, taking advantage of versatile plasmonic heating schemes by varying optical degrees of freedom, such as light polarisation, power, beam position, and short laser pulses.
