Biopolymer-Based Hyperstable Multi-Layered Artificial Synapse with Ultra-Low Energy Consumption

Recently, artificial synapses have attracted significant interest for mimicking the synaptic characteristics of the biological neural system, which are crucial for learning processes in neuroprosthetics, brain-machine interfaces, and deep learning technology. By emulating the neurotransmitter-based biological signal transmission system, artificial synapses have been achieved by integrating multi-terminal transistors or two-terminal memristors, which can conduct synaptic characteristics <i>via</i> migration and accumulation of ions in response to various external stimuli including electric fields, pressure, and light. Multi-terminal transistors in three-electrode systems (gate, source, and drain) obtain synaptic characteristics by regulating the conductance of ionic channels between source and drain electrodes, resulting in non-volatile memory. Because of their structural similarity with biological neural systems in terms of ion migration from source to drain, transistor systems have been widely applied in the development of artificial synapses. However, transistor structures require complicated fabrication processes and consume higher energy than biological synapses owing to the necessity of a gate electrode. Alternatively, two-terminal memristors in metal-resistive layer-metal structures achieve synaptic characteristics by regulating ion flux between metals. Memristor structures offer advantages such as simple device structure with facile fabrication compared to multi-terminal transistors, and low energy consumption owing to robust resistance tunability, enabled by enhanced ion conductivity through ion-doped semiconductors or polymers. Meanwhile, the necessity for developing biopolymers has been addressed owing to the increasing disposal of electronic devices (E-waste) containing plastics and toxic metals, which cause environmental pollution and ecocide, posing threats to both humans and nature. Despite the advantage of biopolymers, such as their abundance in nature and hydrophilic functional groups that facilitate ion migration, their susceptibility to moisture and heat has hindered their utilization in electronic devices.<br/>Here, we develop a biopolymer-based hyperstable and low-energy modulated multi-layered artificial synapse (M-AS) based on a two-terminal structure. The artificial synapse is constructed with a stacked ion active layer (IAL)-ion binding layer (IBL)-ion active layer (IAL) structure, where IALs are crosslinked chitosan (CS) and guar gum (GG) double network with NaCl ions, and the IBL is a dielectric cellulose acetate (CA) that provides ion accumulating sites through ion-dipole coupling (IDC). Since CS and GG polymers assigned their hydrogen bonding to crosslinking, crosslinked CS/GG is less susceptible to moisture, achieving hyperstability. Additionally, IDC at the interface of IAL-IBL facilitates ion accumulation to generate synaptic characteristics including paired-pulse facilitation (PPF) and long-term plasticity (LTP), which can be improved by introducing CA with high dielectric properties. Especially, due to the utilization of highly mobile ions (NaCl) into the IALs to function as neurotransmitters, an extremely low threshold voltage (20 μV) is enough to operate M-AS with remarkably low energy consumption (0.85 fJ), which is lower than that of a biological synapse. Finally, our multi-layered artificial synapse demonstrates an artificial injury system responsive to temperature changes by integrating a thermistor as a mechanoreceptor and robotics to emulate biological responses to injuries such as burns and frostbite. This system effectively alerts to injuries via learning processes, with the severity of injury indicated by the number of activated LEDs and the healing process indicated by the retention of LED illustration.