MRS Meetings and Events


EL20.08.01 2023 MRS Fall Meeting

Describing the Analog Resistance Change of HfOx Neuromorphic Synapses

When and Where

Nov 30, 2023
8:45am - 9:00am

Hynes, Level 3, Room 301



Fabia Farlin Athena1,Eric Vogel1

Georgia Institute of Technology1


Fabia Farlin Athena1,Eric Vogel1

Georgia Institute of Technology1
At the forefront of memristive device research, significant attention is paid to the understanding of resistance variations in Titanium/Hafnium Oxide (Ti/HfO<sub>x</sub>) synapses. These changes are typically dictated by a thin-oxide barrier, a result of oxidation and reduction within a Hafnium-rich conducting filament [1, 2, 3, 4]. Yet, despite a broad understanding of these principles, direct experimental investigation of the conducting filament proves notably complex. Consequently, crucial physical attributes and processes, such as the exact location of the barrier, the temporal fluctuations in thickness during analog pulsing, and the influence of temperature on the current, remain elusive and demand a more thorough elucidation.<br/><br/>We aim to take a step forward to unravel these intricacies by utilizing a compact model rooted in Trap-Assisted-Tunneling and Ohmic transport principles. This model has demonstrated significant value in shedding light on the mechanisms of analog switching in HfO<sub>x</sub> synapses. It not only accurately mirrors the experimentally observed current-voltage relationship, but also faithfully reproduces its sensitivity to temperature changes.<br/><br/>In deploying this model, we have been successful in extracting the barrier heights during analog pulsing. These data align with a barrier situated adjacent to the reset anode [5, 6]. Our findings are supported by an independent Finite Element Analysis simulation using COMSOL Multiphysics<sup>®</sup> that incorporates the migration of oxygen vacancies.<br/><br/>The proposed model's utility has been further amplified by its capacity to estimate the barrier's thickness in response to the analog pulses. It is able to simulate the evolution of the current-voltage relationship from the low resistance state to the high resistance state by simply modulating the thickness of the barrier.<br/><br/>In summary, this work enables us to delve deeper into the relationship between the physical characteristics of the HfO<sub>x</sub> synapses and the analog switching dynamics.<br/><br/>References:<br/>[1] Athena, Fabia F. et al. “Trade-off between Gradual Set and On/Off Ratio in HfO<sub>x</sub>-Based Analog Memory with a Thin SiO<sub>x</sub> Barrier Layer.” <i>ACS Applied Electronic Materials</i> (2023): n. Pag.<br/>[2] Athena, Fabia F., et al. "Towards a better understanding of the forming and resistive switching behavior of Ti-doped HfO<sub>x</sub> RRAM." <i>Journal of Materials Chemistry C</i> 10.15 (2022): 5896-5904.<br/>[3] Athena, Fabia F., et al. "Impact of titanium doping and pulsing conditions on the analog temporal response of hafnium oxide based memristor synapses." <i>Journal of Applied Physics</i> 131.20 (2022): 204901.<br/>[4] West, Matthew P., Athena, Fabia F., Graham, Samuel, Vogel, Eric M. "Bias history impacts the analog resistance change of HfO<sub>x</sub>-based neuromorphic synapses." <i>Applied Physics Letters</i> 122.6 (2023).<br/>[5] Bersuker, G., et al. "Metal oxide RRAM switching mechanism based on conductive filament microscopic properties." <i>2010 International Electron Devices Meeting</i>. IEEE, 2010.<br/>[6] Padovani, Andrea, et al. "Microscopic modeling of HfO<sub>x</sub> RRAM operations: From forming to switching." <i>IEEE Transactions on electron devices</i> 62.6 (2015): 1998-2006.<br/><br/>Acknowledgment:<br/>The authors would like to thank Professor Samuel Graham at the University of Maryland for the COMSOL model and Professor Suman Datta at the Georgia Institute of Technology for providing the thermal measurement facilities.


atomic layer deposition

Symposium Organizers

Gina Adam, George Washington University
Sayani Majumdar, Tampere University
Radu Sporea, University of Surrey
Yiyang Li, University of Michigan

Symposium Support

APL Machine Learning | AIP Publishing

Publishing Alliance

MRS publishes with Springer Nature