Spin-orbit torque driven field-free artificial synapses and neurons
PN-III-P4-ID-PCE-2020-1853 No. 182/04/01/2021
        Ro |En
 

Home TeamResults

 

 

The artificial intelligence (AI) computing systems that can engage in human-like processes are at the forefront of the next technological revolution that will influence most aspects of society. Training an AI emulated on conventional von Neuman computers is an energy intensive process. Therefore, optimizing the power consumption is a prerequisite for the sustainable development of the AI industry. One method to improve the power consumption by orders of magnitude is to use dedicated hardware for neuromorphic computing, like the Spiking Neural Networks (SNNs). Artificial synapses and neurons are at the core of SNNs. Within this project we target to demonstrate viable field-free spin-orbit torque (SOT) driven domain wall (DW) synaptic devices and neurons. The envisioned synaptic device will meet the requirements of spike-timing dependent plasticity, will be energetically efficient, will have superior endurance and a reduced complexity. The SOT driven DW neuron will possess the leaky-integrate-and-fire functionality. The implementation of the project relies on a wide range of techniques starting form thin films elaboration to micro-patterning of functional spin-orbitronic devices.

 

   

 

 
Home | Team | Results