Accelerated Ionic Motion in Amorphous Memristor Oxides for Nonvolatile Memories and Neuromorphic Computing

Abstract

Memristive devices based on mixed ionic–electronic resistive switches have an enormous potential to replace today's transistor‐based memories and Von Neumann computing architectures thanks to their ability for nonvolatile information storage and neuromorphic computing. It still remains unclear however how ionic carriers are propagated in amorphous oxide films at high local electric fields. By using memristive model devices based on LaFeO3 with either amorphous or epitaxial nanostructures, we engineer the structural local bonding units and increase the oxygen‐ionic diffusion coefficient by one order of magnitude for the amorphous oxide, affecting the resistive switching operation. We show that only devices based on amorphous LaFeO3 films reveal memristive behavior due to their increased oxygen vacancy concentration. We achieved stable resistive switching with switching times down to microseconds and confirm that it is predominantly the oxygen‐ionic diffusion character and not electronic defect state changes that modulate the resistive switching device response. Ultimately, these results show that the local arrangement of structural bonding units in amorphous perovskite films at room temperature can be used to largely tune the oxygen vacancy (defect) kinetics for resistive switches (memristors) that are both theoretically challenging to predict and promising for future memory and neuromorphic computing applications.

Flexible Pyrene/Phenanthro[9,10‐d]imidazole‐Based Memristive Devices for Mimicking Synaptic Plasticity

“…By far, memristor, ferroelectric memory, phase‐change memory, resistive memory device, field‐effect transistor, as well as the silicon‐based complementary metal–oxide–semiconductor (CMOS) circuit have been vigorously studied for realization of an ideal synaptic unit. However, device‐level obstacles and barriers still exist.…”

Lithium‐Battery Anode Gains Additional Functionality for Neuromorphic Computing through Metal–Insulator Phase Separation

“…The largest two categories of memristive devices—valence change memories and electrochemical metallization memories—rely on creation, annihilation, and motion of ions and defects through ionic conductors under a bias (which is used to control resistive state). Valence change memories are based on transition metal oxides whose resistive state depends on the redistribution of bulk O 2– and O 2– vacancies under bias (and resulting reduced metal cations) . Typical diffusivity values for state‐of‐the‐art switching oxides (SrTiO 3 , TiO 2 and TaO x ) are 10 −15 –10 −13 cm 2 s −1 at ambient temperature.…”

A Robust and Low‐Power Bismuth Doped Tin Oxide Memristor Derived from Coaxial Conductive Filaments

“…[10,11] Binary metal oxide based memristors are promising for their simple structure, easy control of material composition, compatibility with complementary metal oxide semiconductor technology and cost reduction during the production of hardware components. [12,13] WO x , [14] HfO 2 [7] and TaO x [15,16] have been widely studied due to their attractive performance. Tin oxide (SnO 2 ) as a binary semiconductor, has been widely applied for transparent sensors, [17,18] solar cells, [19,20] and photocatalysis.…”