Solution-Processed Polymer Memcapacitors with Stimulus-Controlled and Evolvable Synaptic Functionalities: From...

Area of Science:

• Neuromorphic Engineering
• Materials Science
• Neuroscience

Background:

• The human brain's ability to learn and adapt relies on synaptic plasticity.
• Developing artificial systems that replicate this flexibility is a key goal in neuromorphic engineering.
• Existing artificial synapses often lack the dynamic range and adaptability of biological counterparts.

Purpose of the Study:

• To develop a novel paradigm of biocompatible polymer memcapacitors.
• To demonstrate comprehensive synaptic capabilities, including short-term plasticity (STP), long-term plasticity (LTP), and metaplasticity (MP).
• To explore the application of these memcapacitors in artificial neural networks with dynamic learning rates.

Main Methods:

• Fabrication of biocompatible polymer memcapacitors via a seamless solution process.
• Characterization of memcapacitive behavior under varying stimulation frequencies and intensities.
• Investigation of stimulus-controlled spatiotemporal ion redistribution within the polymer.
• Implementation of memcapacitors with dynamic learning rates in an artificial neural network.

Main Results:

• Memcapacitors exhibited analogue-type and evolvable capacitance shifts, mimicking synaptic strengthening and weakening.
• Demonstrated a transition from STP to LTP and further to MP with increasing stimulation.
• Elucidated the physical mechanism of ion redistribution responsible for versatile synaptic plasticity.
• Showcased the superiority of dynamic learning rates over constant rates in an artificial neural network.

Conclusions:

• The developed polymer memcapacitors offer a promising platform for organic neuromorphic computing.
• The demonstrated metaplasticity enables dynamic learning rate adaptation, enhancing artificial neural network performance.
• This work advances the field of biocompatible neuromorphic devices with brain-like learning capabilities.