Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2021 Aug 9;11(1):16094.
doi: 10.1038/s41598-021-95569-4.

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

Affiliations

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

Leandro Martins et al. Sci Rep. .

Abstract

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.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
(a) Top view optical image of the devices after fabrication with an integrated local field line fabricated on top of the MTJ nanopillar. (b) Illustration of the experimental setup used for the initialization and spin diode measurements. (c) Two terminal R-H loop for an MTJ with a diameter of 1.0 µm, explaining the 3-step initialization process. The insets show how each chirality is associated with a different vortex configuration of the in-plane magnetic moments of the free layer, as seen from the bottom of the MTJ.
Figure 2
Figure 2
Spin diode measurements presented as a function of the input RF current density for (a) positive and (b) negative chiralities. (c) Spin diode measurement obtained for an input RF current density of 2.8 × 109 A/m2. (d) Rectified voltage measured at a fixed frequency of 60 MHz and presented as a function of the input RF current density.
Figure 3
Figure 3
(a) Simulated rectified voltage calculated as a function of the input RF current density for the two different vortex chiralities at a fixed frequency of 81 MHz. An in-plane magnetic field of 39.8 Oe is applied in the yy direction. The simulated spin diode measurements for (b) positive and (c) negative chiralities are presented for different excitation mechanisms: spin-transfer torque (STT), Oersted field torque (OFT) and where both mechanisms are considered (OFT + STT). The results were obtained for an RF current density of 3.0 × 109 A/m2. (d,e) Simulated ΔMY time traces obtained for both chiralities. Similar to (b) and (c), different excitation mechanisms are considered. The results were obtained for an RF current density of 3.0 × 109 A/m2 at a frequency of 81 MHz whose time traces are also presented.
Figure 4
Figure 4
(a) Difference of the output voltage (ΔV) calculated for C =  + 1 and C = − 1. For each setpoint field, ΔV is calculated at the frequency fMAX that maximizes the difference. (b) Example of the ΔV calculation for a setpoint field of 18.0 Oe.
Figure 5
Figure 5
Spin diode measurements obtained for an STVO with a diameter of (a) 1.0 µm and (b) 0.9 µm, for an RF input power of 0.398 mW. The four possible programmable sums of the output voltages are presented in (c) as a function of the input RF power for a frequency of 60 MHz.

References

    1. Krizhevsky A, Sutskever I, Hinton G. ImageNet classification with deep convolutional neural networks. Proc. Adv. Neural Inf. Process. Syst. 2012;25:1090–1098.
    1. De Santana LMQ, Santos RM, Matos LN, Macedo HT. Deep neural networks for acoustic modeling in the presence of noise. IEEE Lat. Am. Trans. 2018;16:918–925. doi: 10.1109/TLA.2018.8358674. - DOI
    1. Xiong HY, et al. The human splicing code reveals new insights into the genetic determinants of disease. Science. 2015;347:1254806. doi: 10.1126/science.1254806. - DOI - PMC - PubMed
    1. McMahon P, et al. A fully programmable 100-spin coherent Ising machine with all-to-all connections. Science. 2016;354:614–617. doi: 10.1126/science.aah5178. - DOI - PubMed
    1. Merolla PA, et al. Amillion spiking-neuron integrated circuit with a scalable communication network and interface. Science. 2014;345:668–673. doi: 10.1126/science.1254642. - DOI - PubMed

Publication types