Ferroelectric Tunnel Junctions for neuromorphic applications

The recent discovery of HfO2 ferroelectric properties generates a strong interest for novel non-volatile memory technologies. Among them, HfO2-based Ferroelectric Tunnel Junctions (FTJ) are resistive memories in which the electronic transport, and thus the device conductance, is modulated by the orientation of ferroelectric dipoles within the HfO2 layer. Actually, HfO2-based FTJs are envisaged for mimicking human brain synapses in neuromorphic circuits thanks to their ability of progressive electrical programming (potentiation, depression…), together with non-destructive reading of multi-states operations.

The PhD will consist in:

– Optimizing single FTJ bitcells towards fast read operations: this will consist in scaling down the ferroelectric layer thickness below 10nm, which is expected to be detrimental for FTJ operations. Achieving thinner ferroelectric layers while preserving a satisfactory conductance modulation together with multilevel capability is a challenge, which will require a fine understanding of ferroelectric material at crystallographic level.

– Assessing the performance of FTJ technology for neuromorphic applications, by means of dedicated electrical sequences mimicking synapse signals.

– Characterizing FTJ devices in realistic circuit environment, taking advantage of dedicated neuronal designs

Facebook

Twitter