Perpendicular-anisotropy artificial spin ice with spontaneous ordering: a platform for neuromorphic computing with flexible timescales

Aleksandr Kurenkov,Jonathan Maes,Aleksandra Pac,Gavin Martin Macauley,Bartel Van Waeyenberge,Aleš Hrabec,Laura Jane Heyderman

2024-08-22

Abstract:Arrays of coupled nanomagnets have wide-ranging fundamental and practical applications in artificial spin ices, neuromorphic computing and spintronics. However, lacking in these fields are nanomagnets with perpendicular magnetic anisotropy with sufficient magnetostatic interaction. This would not only open up new possibilities for artificial spin ice geometries but also enable novel coupling methods for applications. Here we demonstrate a method to engineer the energy landscape of artificial spin lattices with perpendicular magnetic anisotropy. With this, we are able to realize for the first time magnetostatically-coupled 2D lattices of out-of-plane Ising spins that spontaneously order at room temperature on timescales that can be precisely engineered. We show how this property, together with straightforward electrical interfacing, make this system a promising platform for reservoir computing. Our results open the way to investigate the thermodynamics of out-of-plane magnetostatically coupled nanomagnet arrays with novel spin ice geometries, as well as to exploit such nanomagnet arrays in unconventional computing, taking advantage of the adjustable temporal dynamics and strong coupling between nanomagnets.

Mesoscale and Nanoscale Physics,Computational Physics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to realize an array of nanomagnets with perpendicular magnetic anisotropy in the artificial spin ice system, and make them spontaneously and orderly arranged at room temperature, and this kind of arrangement can be achieved on an experimentally measurable time scale. Specifically, the paper aims to:

1. **Develop nanomagnets with perpendicular magnetic anisotropy**: Traditional nanomagnets usually have in - plane magnetic anisotropy, which limits their use in some applications. Through designing a specific multi - layer thin - film structure, the paper realizes nanomagnets with perpendicular magnetic anisotropy.

2. **Achieve spontaneous and orderly arrangement**: The paper shows how to make these nanomagnets spontaneously and orderly arranged at room temperature by precisely controlling the size and spacing of nanomagnets as well as the number of layers and thickness of the multi - layer thin - film. This spontaneous and orderly arrangement is achieved by adjusting the energy landscape of the system, so that the magnetostatic coupling between nanomagnets is strong enough to promote the orderly arrangement.

3. **Control the time scale**: The paper also studies how to precisely control the relaxation time scale of the nanomagnet array by adjusting system parameters (such as the effective anisotropy energy \(E_{\text{EA}}\) and the magnetostatic coupling energy \(E_{\text{MC}}\)). This enables the system to achieve an orderly arrangement on different time scales from sub - seconds to years, thus providing flexibility for different application requirements.

4. **Application prospects**: The paper explores the potential applications of this kind of nanomagnet array with perpendicular magnetic anisotropy in fields such as neuromorphic computing and spintronics. In particular, this system can be used as a platform for reservoir computing, taking advantage of its adjustable dynamic characteristics and strong coupling characteristics.

### Formula Explanation

- **Néel - Arrhenius Law**:
\[
\tau_j(t)=\tau_0\exp\left(\frac{\Delta E_j(t)}{k_BT}\right)
\]
where:
- \(\tau_j(t)\) is the switching time of the \(j\) - th nanomagnet at time \(t\).
- \(\tau_0 = 10^{- 10}\text{s}\) is the attempt period.
- \(\Delta E_j(t)\) is the energy barrier of the \(j\) - th nanomagnet at time \(t\).
- \(k_B\) is the Boltzmann constant.
- \(T\) is the temperature.

- **Energy Barrier**:
\[
\Delta E_j(t)=E_{\text{EA}}(1 + \text{rand}(j))+E_{\text{MC}}\sum_{i,j}S_i(t)D_{ij}S_j(t)
\]
where:
- \(E_{\text{EA}}\) is the effective anisotropy energy.
- \(\text{rand}(j)\) is a random Gaussian variation with a range of ±5%.
- \(E_{\text{MC}}\) is the magnetostatic coupling energy.
- \(S_i(t)\) and \(S_j(t)\) are the spin states of the \(i\) - th and \(j\) - th nanomagnets at time \(t\), respectively.
- \(D_{ij}\) is the strength of the point - dipole interaction between the \(i\) - th and \(j\) - th nanomagnets.

Through these studies, the paper not only solves the technical challenges, but also provides new possibilities for future applications in fields such as neuromorphic computing and spintronics.