Improved synaptic properties of HfSiO_x-based ferroelectric memristors by optimizing Ti/N ratio in TiN top electrode for neuromorphic computing

Recently, hafnium oxide-based ferroelectric memristors have attracted considerable research interest because they offer several advantages over conventional perovskite-based ferroelectric memristors, including better CMOS compatibility, lower power consumption, and enhanced scalability. HfO_x doped with silicon (Si) shows great potential, as silicon’s smaller atomic radius compared with hafnium enhances the induction of ferroelectric properties. This study focuses on HfSiO_x-based ferroelectric memristors, examining the electrical characteristics of the devices by varying the DC power during the sputtering of the TiN top electrode while maintaining consistent silicon doping and material composition. Moreover, we emphasize the relationship between DC power and endurance, and assess the electrical characteristics of the devices by evaluating maximum remnant polarization (2P_r), interfacial capacitance (C_i), and tunneling electro resistance (TER) under various DC power conditions. Additionally, by investigating synaptic properties of ferroelectric memristors through potentiation and depression (P&D), and paired-pulse facilitation (PPF), the study demonstrates their potential for applications in neuromorphic computing.