Using honey-CNT memristors in neuromorphic computing
DOI: 10.1063/10.0023966
Using honey-CNT memristors in neuromorphic computing lead image
Neuromorphic computing is an emerging computing technology inspired by systems inside the human brain. One of its necessary hardware components is a memristor; its metal-dielectric-metal device structure models the presynapse-synaptic cleft-postsynapse structure of the biological neuron, allowing it to emulate neuronal and synaptic functions.
In recent years, the growing need to reduce electronic waste has prompted a search for natural biomaterials that can be used as the dielectric layer of memristors. One such biomaterial is honey. In order to enhance its viability, Templin et al. incorporated carbon nanotubes (CNTs) in honey and tested the device’s ability to reproduce synaptic plasticity.
“Natural materials previously found to exhibit memristive properties include sugars and proteins,
including the honey that we have been investigating,” said author Feng Zhao. “By incorporating carbon nanotubes in the honey film, our honey memristor shows highly reproducible nonvolatile memory behaviors and capability of emulating a variety of synaptic functions.”
The authors tested the honey-CNT device’s ability to achieve spike-timing-dependent plasticity, paired-pulse facilitation, and long term and long-term memory. Their results show that neuromorphic computing systems employing honey-CNT memristors are both efficient and sustainable.
In the future, the authors plan to explore honey memristors at the nanometer scale.
“Honey memristors can provide energy efficient operation with improved sustainability and eco-friendly disposal,” said Zhao. “Currently, we are investigating approaches to scale up honey memristors with nanometer size and high density in order to mimic nervous systems and neural networks for neuromorphic systems.”
Source: “Synaptic plasticity emulation by natural biomaterial honey-CNT-based memristors,” by Zoe Templin, Md Mehedi Hasan Tanim, and Feng Zhao, Applied Physics Letters (2023). The article can be accessed at https://doi.org/10.1063/5.0174426 .
