Abstract Body: The rapid increase in data quantity and processing necessitate novel approaches to computing. Digital computing based on von Neumann architectures may not be optimal for data-intensive operations like machine learning due to the high energy costs of moving data between processor and memory. Neuromorphic computing that combines logic and memory functionality into a single analogue nonvolatile memory element is a promising energy-efficient solution that decreases the energy consumption by orders of magnitude. Filament-based phase-change memory and resistive random access memory are the most well-investigated approaches to neuromorphic computing, but these memories are stochastic due to the unreliability of the nanosized filaments used to store information. In this work, we develop the “bulk” resistive random access memory that stores information using the oxygen vacancy concentration in the 3D bulk. This cell contains a solid electrolyte sandwiched between two mixed ionic and electronic conductors (MIECs). The ion-conducting solid electrolyte blocks electrons and prevents the positive feedback that results in filament formation. Instead, the oxygen vacancy ions are uniformly distributed in the absence of filaments and thermal gradients. To switch the cell, electrochemical voltage pulses (+/-1.5V, 2 μs) shuttle vacancies between the MIECs; the concentration of the vacancies in one MIEC stores the analogue resistive state. This enables exceptionally reliable, linear, and deterministic switching among hundreds of analogue information states, while utilizing CMOS-compatible materials with long endurance and information retention. Atomic force microscopy and physical modeling confirms the absence of filaments. Three types of solid electrolytes are utilized: a thick model single-crystal yttrium-stabilized zirconia (YSZ), a thin-film YSZ grown by sputtering, and ultra-thin-film electrolytes of alumina, hafnia, and zirconia grown by atomic layer deposition. The oxygen vacancy ionic conductivities of these materials play a crucial role in the switching and retention times. The broad classes of oxygen conductors provide significant opportunities for future materials research towards next-generation analogue nonvolatile memory. Reference: “Filament-free bulk resistive memory enables deterministic analogue switching,” Adv Mater, 32, 2003984 (2020)

Deterministic Analogue Resistive Memory for Neuromorphic Computing Based on Transition Metal Oxides

DD: Materials for Memory and Logic

technical paper

The Electronic Materials Conference (EMC) is the premier annual forum on the preparation, characterization and use of electronic materials. This year’s event will feature a plenary session, topical sessions, a poster session and more.

EMC is coordinated with the Device Research Conference (DRC), which will be held during the same week, from June 20-23. This recognizes the strong interaction between device and electronic materials research and provides fruitful exchange of information between attendees of both Conferences.

![](https://assets.underline.io/uploads/markdown_image/1/image/eb26c24a18d5d1030616afa4f993cf89.png)
	
<a href="https://www.mrs.org/docs/default-source/meetings-events/mrs-conference-services/2021/emc-2021/2021-emc-program-at-a-glance-6-16-21.pdf?sfvrsn=e39f950c_4">Latest Program-at-a-Glance</a>

<a href="https://drive.google.com/file/d/1eiS1ao4p83VNy8xgHs3Rls7gTQjx7Rc4/view?usp=sharing">Full Program with Abstracts</a>

Diana Kim

1

3

SHORT BIO

Presentations

Deterministic Analogue Resistive Memory for Neuromorphic Computing Based on Transition Metal Oxides