Abstract
Advancing artificial neural networks requires the replication of multiple biological features to handle complex tasks in dynamic working environments. Oxide-based resistive memories, with superior uniformity, faster switching speeds, and reduced device dimensions, surpass traditional complementary metal–oxide–semiconductor (CMOS) technology for neural networks. To address the limitations of non-volatile memristors, particularly their large performance variations, this study introduces a TiN/GaOx/Pt resistive-switching device. Endurance and retention tests confirm the device's stability and uniformity, while its ability to replicate key biological functions is demonstrated through synaptic and nociceptive behaviors. By modulating synaptic plasticity under the Hebbian learning rule, the device mimics excitatory postsynaptic current (EPSC) and spike time-dependent plasticity (STDP). Additionally, it exhibits nociceptor traits by generating current responses aligned with various pulse-configured inputs. This novel memristor marks a significant advancement in bioinspired technology, enabling the simultaneous emulation of biological nociceptors and synapses, and paving the way for next-generation artificial neural networks and humanoid robotics.
| Original language | English |
|---|---|
| Article number | 162973 |
| Journal | Applied Surface Science |
| Volume | 697 |
| DOIs | |
| State | Published - 15 Jul 2025 |
Keywords
- Artificial nociceptor
- Artificial synapse
- Non-volatile memory
- Resistive switching device
Fingerprint
Dive into the research topics of 'Emulating nociceptor and synaptic functions in GaOx-based resistive random-access memory for bio-inspired computing'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver