Interface-engineered reliable HfO2-based RRAM for synaptic simulation

Future synaptic simulation using resistance random access memory (RRAM) requires higher reliability and lower power consumption of the devices and understanding of the correlation of the materials with their multi-level resistance switching (RS) properties. Using O3 pretreatment on a TiN electrode,...

Full description

Saved in:
Bibliographic Details
Published in:Journal of materials chemistry. C, Materials for optical and electronic devices Materials for optical and electronic devices, 2019, Vol.7 (40), p.12682-12687
Main Authors: Wang, Qiang, Niu, Gang, Roy, Sourav, Wang, Yankun, Zhang, Yijun, Wu, Heping, Zhai, Shijie, Bai, Wei, Shi, Peng, Song, Sannian, Song, Zhitang, Ya-Hong, Xie, Zuo-Guang Ye, Wenger, Christian, Meng, Xiangjian, Ren, Wei
Format: Article
Language:English
Subjects:
Devices
Hafnium oxide
Insulators
Photoelectrons
Power consumption
Pretreatment
Properties (attributes)
Random access memory
Reliability
Simulation
Synapses
Titanium dioxide
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Future synaptic simulation using resistance random access memory (RRAM) requires higher reliability and lower power consumption of the devices and understanding of the correlation of the materials with their multi-level resistance switching (RS) properties. Using O3 pretreatment on a TiN electrode, this work highlights the significant role of the interface in the enhancement of the reliability and the power consumption of HfO2-based RRAM devices. X-ray photoelectron spectroscopy investigations indicate increases of the TiON and TiO2 components with the augmentation of the number of O3 treatment cycles, which strongly impacts the RS properties of the Pt/HfO2/TiN devices. Optimal RS properties were obtained for 20 O3 pulse-pretreated devices, which were used to emulate biological synapses after an annealing process. Analog memory properties, including analog set and reset in DC mode and potentiation/depression based on two types of designed pulses, have been achieved. Finally, one of the biological synapse learning rules, spike-timing-dependent plasticity, was successfully emulated. These results, avoiding the conventional route based on dual-layer insulators, are of significance for synaptic simulation using interface-engineered single-layer HfO2 RRAM and further reveal the internal mechanism of HfO2-based electron synapses.
ISSN:2050-7526
2050-7534
DOI:10.1039/c9tc04880d