Microstructural implications for neuromorphic synapses based on ferroelectric hafnium oxide

Abstract

Recently, ferroelectric field-effect transistors (Fe-FETs) based on hafnium oxide (HfO2) have been shown to be promising candidates for synaptic devices in neuromorphic applications. The polycrystalline structure of the ferroelectric layer strongly impacts the memory storage as well as the synaptic device performance, especially for highly scaled FeFETs. Therefore, detailed understanding of the transition behavior for synaptic potentiation and depression cases is essential. Combining the measurement results of grain structure, crystallographic phases and texture obtained by means of transmission Kikuchi diffraction (TKD) with a current percolation path (CPP) simulation model scaling trends for FeFETs suitable for neuromorphic applications can be explained. Results show that after overcoming a grain size dependent minimum channel length (L) width (W) scaling is the key factor to control the switching transition slope.