NeuroImprint

Bioinspired neuromorphic devices have attracted considerable attention due to their potential application in neuro-robotics or neuro-prosthetics. There are two fundamental advantages of neuromorphic devices. First, when mimicking neural and synaptic functionalities, processing and storage are performed physically at the same place, so that they can overcome the well-known von Neumann bottleneck that currently limits the speed and efficiency of computational architectures due to the spatial separation of computation and storage units. Second, if digital Si-technology is applied to merely emulate synapses or nerves, a large number of devices is required due to the analog-to- digital conversion, whereas biological synapses work in analog modes, requiring far fewer devices. Neurons and synapses in a brain perform memory and processing in an integrated manner, which is the essence of associative learning, and they function energy-efficiently by analog adjustment of synaptic weights in response to stimulation. This inspired the development of neuromorphic devices and systems that mimic these characteristic functions of neurons and synapses. Organic materials devices are good candidates for such systems and could provide advantages of biocompatibility, low cost, low energy switching, low working voltage, excellent tunability and provide good mimic of the functions of biological synapses. Recently, bioinspired interactive neuromorphic devices with the ability to directly sense/store/process various stimuli from external environment have been realized by integrating sensors with synaptic devices on flexible substrates. The key remaining challenges for these systems include device reproducibility, limited performance and the absence of a fully scalable fabrication process. Based on large-area compatible direct 3D nanoimprinting processes, which we call DINOFED, we plan to fabricate innovative organic transistors and circuits with minimized critical dimensions on a single substrate. The proposed DINOFED process fulfills all requirements for the next generation flexible electronics, being large-area, parallel patterned with high throughput, high-resolution and inherently self-aligned. By integrating the nanoimprinted transistors and circuits with a tactile sensor on a single conformable substrate, an artificial sensory neuron is realized and tested as neuromorphic learning device with the ability to perceive and memorize physical inputs.