Voltage-Triggered Emergent Dynamics in Strongly Coupled Nanomagnet Networks for Neuromorphic Computing.

Abstract

Emergent dynamics, which arise from local interactions between many elementary components, are central to complex physical and biological systems. However, realizing such dynamics in artificial materials, particularly under low-energy stimuli, remains a challenge. While dipole-dipole interactions are typically suppressed in magnetic storage, here we amplify and use them as a core mechanism to construct a strongly dipolar-coupled network of SmCo₅ macrospins at the wafer scale, which can exhibit intrinsic interaction-driven collective dynamics in response to voltage pulses. This network integrates three key ingredients: the strong dipole-dipole interaction, giant voltage control of coercivity over nearly 1000-fold, and a network topology with a frustrated Ising-like energy landscape. As a result, the network, when stimulated by ∼1 V pulses, transitions from a high-coercivity memory regime into a low-coercivity regime in which internal dipolar fields alone trigger collective magnetic reconfiguration. In this regime, the network exhibits emergent behaviors absent at the level of isolated macrospins, including spontaneous demagnetization, greatly enhanced magnetization modulation, reversible "freeze and resume" evolution, and stochastic convergence toward low-energy magnetic configurations. Furthermore, as a conceptual illustration, micromagnetic simulation of such a strongly dipolar-coupled network shows that the resulting high-dimensional collective dynamics can support temporal information processing, such as accurate chaotic Mackey-Glass prediction and multiclass drone-signal classification. Our work suggests a conceptually distinct route toward scalable, energy-efficient neuromorphic computing, one rooted in local physical interaction-driven emergent dynamics at the network level rather than merely mimicking individual neurons and synapses.