Research articlesSynchronization properties and reservoir computing capability of hexagonal spintronic oscillator arrays
Introduction
Spintronic oscillators have been drawing a lot of attention as devices that could be utilized for radio-frequency signal generation and transmission as well as for bio-inspired neuromorphic computing [1], [2], [3], [4], [5], [6], [7]. Multiple physical implementations, including spin-torque nano-oscillators [8], [9], [10], [2], [6], [7],spin Hall nano-oscillators [11], [12], [2], [13], oscillators based on antiferromagnets [14], [15], [16] or uncompensated ferrimagnets [17] have been proposed, featuring advantages such as the ability to work in room temperature, energy efficiency, small footprint and high frequency ranges being available [1], [2], [3]. The common thread that remains the key to both of the main application areas, namely the DC-to-RF signal conversion and unconventional computing, is making use of oscillator networks and their ability to synchronize by mutually influencing each oscillator’s dynamics [1], [2], [3], [18], [19], [20]. Therefore, a thorough understanding of the coupling behavior is expected to be crucial for further advancing the spintronic oscillator research.
At the same time, synchronization theory in complex networks of oscillators has seen substantial progress within the last several years, creating solid theoretical foundations for understanding most dynamical phenomena occurring in these systems [21], [22], [23]. In this approach, a single oscillator is usually considered to be an abstract object characterized only by its phase, natural frequency and the coupling presence, leading to a simple dynamics equation known as Kuramoto equation [24], [21]. Recently, a number of works have shown that Kuramoto equation can be successfully used to describe arrays of spintronic oscillators as well, both in its basic form and in the modified form where the phase degree of freedom is replaced by two degrees of freedom corresponding to oscillation phase and amplitude or power [25], [26], [20]. Typically, such a description assumes a coupling form that scales with distance as , which is characteristic to magnetodipolar coupling that naturally occurs in physical devices such as spin-torque oscillators or spin Hall oscillators [26], [20], [27].
There are, however, important barriers that often make it difficult to directly apply the findings of the oscillation networks theory in the applied spintronics research. Most notably, there exists a vast number of different Kuramoto equation models, from which only a relatively small number can be considered immediately relevant to the spintronic oscillator landscape: many of the commonly made assumptions, such as using periodic boundary conditions, arbitrarily rewired coupling paths in order to increase disorder or an all-to-all coupling scheme, do not hold for a finite-size, magnetically coupled network [28], [21], [22], [23], [29], [30]. One direction of research that remains relatively unexplored is the influence of grid geometry on synchronization properties. While for an all-to-all coupling scheme the relative physical distances and positions of oscillators can be mostly ignored, in the case of a 2D system with spatially-dependent coupling they can be expected to influence the network dynamics significantly. Despite this, a simple arrangement of oscillators on a cubic Cartesian grid is often assumed in literature by default [31], [32], [30], [26], [20], with the notable exception of the recent works on nanoconstriction-based oscillators, where it was found that the synchronization can be helped by using a tilted rectangular array that assists in spin wave propagation along certain directions. [33], [34].
This work investigates synchronization properties and selected neuromorphic computation capabilities for a 2-D oscillator array on a hexagonal grid and compares the obtained results with a rectangular (cubic) grid. Two important cases are examined: firstly, large-scale systems with periodic boundary conditions and nearest-neighbor coupling are considered, in order to highlight some of the differences between the hexagonal arrangement and the results known from the existing synchronization theory literature. Secondly, a more realistic case of a small-scale network with no periodic boundary conditions and coupling decay is analyzed as a direct model of a spintronic oscillators array with magnetic coupling. The synchronization threshold, emission power and short-memory capabilities are all shown to depend on the grid choice significantly, opening new possibilities for oscillator networks research and applications.
Section snippets
Large-scale systems with periodic boundary conditions
The basic geometrical scheme we examine in this section is pictured in Fig. 1, where nearest-neighbors and second-nearest-neighbors of a single grid point are shown for the rectangular (a) and the hexagonal (b) grid. Taking into consideration only the nearest-neighbors, the Kuramoto equation for the middle oscillator can be written as [21]:in the rectangular case, where is the phase of the oscillator
Small-scale systems
The results derived in the previous section for large-scale systems with periodic boundary conditions indicate different synchronization behavior in hexagonal arrays of nearest-neighbor Kuramoto oscillators when compared to their rectangular counterparts. However, this alone is not sufficient to make a prediction about the behavior of a small-scale system without any periodic boundary conditions, which is often the most interesting case from the application point of view. We choose to focus
Summary
We have demonstrated the influence of grid geometry on selected properties of a 2-D oscillator array with spatially-dependent coupling. By using hexagonal as opposed to cubic lattice for oscillator positions, synchronization threshold was decreased and emission power was increased for both large-scale systems with periodic boundary conditions and for small-scale systems corresponding to a realistic setup of spintronic oscillators such as spin-torque or spin Hall nano-oscillators. Additionally,
CRediT authorship contribution statement
Jakub Chęciński: Conceptualization, Methodology, Software, Validation, Formal analysis, Visualization, Writing - original draft, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
Numerical calculations were supported by PL-GRID infrastructure.
References (38)
- et al.
The kuramoto model in complex networks
Phys. Rep.
(2016) - et al.
Spin-torque building blocks
Nat. Mater.
(2014) - T. Chen, R.K. Dumas, A. Eklund, P.K. Muduli, A. Houshang, A.A. Awad, Ph. Dürrenfeld, B.G. Malm, A. Rusu, J. Åkerman,...
- B. Dieny, I.L. Prejbeanu, K. Garello, P. Gambardella, P. Freitas, R. Lehndorff, W. Raberg, U. Ebels, S.O. Demokritov,...
- H.S. Choi, S.Y. Kang, S.J. Cho, I.-Y. Oh, M. Shin, H. Park, Ch. Jang, B.-Ch. Min, S.-Il Kim, S.-Y. Park, et al., Spin...
- et al.
Frequency shift keying by current modulation in a mtj-based stno with high data rate
Appl. Phys. Lett.
(2017) - et al.
Neuromorphic computing with nanoscale spintronic oscillators
Nature
(2017) - et al.
Vowel recognition with four coupled spin-torque nano-oscillators
Nature
(2018) - et al.
Microwave oscillations of a nanomagnet driven by a spin-polarized current
Nature
(2003) - et al.
Spin torque oscillator frequency versus magnetic field angle: The prospect of operation beyond 65 ghz
Appl. Phys. Lett.
(2009)
Coupled spin torque nano oscillators for low power neural computation
IEEE Trans. Magn.
Magnetic oscillations driven by the spin hall effect in 3-terminal magnetic tunnel junction devices
Phys. Rev. Lett.
Magnetic nano-oscillator driven by pure spin current
Nat. Mater.
Long-range mutual synchronization of spin hall nano-oscillators
Nat. Phys.
Terahertz antiferromagnetic spin hall nano-oscillator
Phys. Rev. Lett.
Antiferromagnetic thz-frequency josephson-like oscillator driven by spin current
Scientific Rep.
Subterahertz ferrimagnetic spin-transfer torque oscillator
Phys. Rev. B
Mutual synchronization of spin torque nano-oscillators through a long-range and tunable electrical coupling scheme
Nat. Commun.
Cited by (9)
Secure Communication via Chaotic Synchronization Based on Reservoir Computing
2024, IEEE Transactions on Neural Networks and Learning SystemsTilted magnetic anisotropy-tailored spin torque nano-oscillators for neuromorphic computing
2023, Applied Physics LettersReservoir Computing with Nanowire Exchange-Coupled Spin Torque Oscillator Arrays
2023, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)Ultra-Low-Power Reservoir Computing Based on Synthetic Antiferromagnetic Skyrmion Pairs
2022, IEEE Electron Device Letters