Imitating human-like learning and text encoding utilizing flexible and sustainable synaptic transistors composed of recyclable materials.

Recycling has become the best solution for addressing ever-rising electronic waste (e-waste) and sustainable waste reduction to protect human and ecological health. The use of recyclable and nature-originated materials can provide an ecofriendly route towards the development of smart electronics for data-intensive requirements. In this work, sustainable synaptic transistors based on recyclable, organic, and natural materials are demonstrated for sustainable neuromorphic systems, especially for excellent human-like learning and text encoding. The fabricated transistors exhibited p-channel characteristics alongside remarkable non-volatile memory behavior. Moreover, in addition to demonstrating superior operational stability, these devices can replicate spike timing-dependent, voltage-dependent, and number-dependent plasticity and pulse-paired facilitation. These devices maintained their ability to mimic even after undergoing extensive flexibility assessments. These transistors can encode text as Morse code while operating at a remarkably low energy per synaptic event of 0.035 fJ. Finally, these devices achieved an accuracy of 81.2% when the extracted weights were applied in a simple three-layer ANN architecture designed for recognizing everyday fashion apparel. The recognition accuracy remained unaltered even after the devices were bent multiple times at various radii. These exciting results suggest a potential new direction for the development of hardware-based sustainable and flexible neuromorphic systems.