Multilayer spintronic neural networks with radiofrequency connections.

Spintronic nano-synapses and nano-neurons perform neural network operations with high accuracy thanks to their rich, reproducible and controllable magnetization dynamics. These dynamical nanodevices could transform artificial intelligence hardware, provided they implement state-of-the-art deep neural networks. However, there is today no scalable way to connect them in multilayers. Here we show that the flagship nano-components of spintronics, magnetic tunnel junctions, can be connected into multilayer neural networks where they implement both synapses and neurons thanks to their magnetization dynamics, and communicate by processing, transmitting and receiving radiofrequency signals. We build a hardware spintronic neural network composed of nine magnetic tunnel junctions connected in two layers, and show that it natively classifies nonlinearly separable radiofrequency inputs with an accuracy of 97.7%. Using physical simulations, we demonstrate that a large network of nanoscale junctions can achieve state-of-the-art identification of drones from their radiofrequency transmissions, without digitization and consuming only a few milliwatts, which constitutes a gain of several orders of magnitude in power consumption compared to currently used techniques. This study lays the foundation for deep, dynamical, spintronic neural networks.

Non-volatile artificial synapse based on a vortex nano-oscillator.

In this work, a new mechanism to combine a non-volatile behaviour with the spin diode detection of a vortex-based spin torque nano-oscillator (STVO) is presented. Experimentally, it is observed that the spin diode response of the oscillator depends on the vortex chirality. Consequently, fixing the frequency of the incoming signal and switching the vortex chirality results in a different rectified voltage. In this way, the chirality can be deterministically controlled via the application of electrical signals injected locally in the device, resulting in a non-volatile control of the output voltage for a given input frequency. Micromagnetic simulations corroborate the experimental results and show the main contribution of the Oersted field created by the input RF current density in defining two distinct spin diode detections for different chiralities. By using two non-identical STVOs, we show how these devices can be used as programmable non-volatile synapses in artificial neural networks.

Digital and analogue modulation and demodulation scheme using vortex-based spin torque nano-oscillators.

In conventional communications systems, information is transmitted by modulating the frequency, amplitude or phase of the carrier signal, which often occurs in a binary fashion over a very narrow bandwidth. Recently, ultra-wideband signal transmission has gained interest for local communications in technologies such as autonomous local sensor networks and on-chip communications, which presents a challenge for conventional electronics. Spin-torque nano-oscillators (STNOs) have been studied as a potentially low power highly tunable frequency source, and in this report we expand on this to show how a specific dynamic phase present in vortex-based STNOs makes them also well suited as Wideband Analogue Dynamic Sensors (WADS). This multi-functionality of the STNOs is the basis of a new modulation and demodulation scheme, where nominally identical devices can be used to transmit information in both a digital or analogue manner, with the potential to allow the highly efficient transmittance of data.

Nanoscale true random bit generator based on magnetic state transitions in magnetic tunnel junctions.

We present an investigation into the in-plane field induced free layer state transitions found in magnetic tunnel junctions. By applying an ac current into an integrated field antenna, the magnetisation of the free layer can be switched between the magnetic vortex state and the quasi-uniform anti-parallel state. When in the magnetic vortex state, the vortex core gyrates a discrete number of times, and at certain frequencies there is a 50% chance of the core gyrating n or n - 1 times, leading to the proposal of a novel nanoscale continuous digital true random bit generator.

Self-Injection Locking of a Vortex Spin Torque Oscillator by Delayed Feedback.

The self-synchronization of spin torque oscillators is investigated experimentally by re-injecting its radiofrequency (rf) current after a certain delay time. We demonstrate that the integrated power and spectral linewidth are improved for optimal delays. Moreover by varying the phase difference between the emitted power and the re-injected one, we find a clear oscillatory dependence on the phase difference with a 2π periodicity of the frequency of the oscillator as well as its power and linewidth. Such periodical behavior within the self-injection regime is well described by the general model of nonlinear auto-oscillators including not only a delayed rf current but also all spin torque forces responsible for the self-synchronization. Our results reveal new approaches for controlling the non-autonomous dynamics of spin torque oscillators, a key issue for rf spintronics applications as well as for the development of neuro-inspired spin-torque oscillators based devices.