Multi-Gate Transistors on Flexible Substrates as Building Blocks for Neuromorphic Electronic Skin

Electronic-skin (e-skin) has garnered considerable interest as a medium that allows robots to feel and perceive the real world, just like biological skin. During these tasks, the e-skin generates large datasets, which need to be handled efficiently and often in real time. Considering this, brain-inspired computing, particularly the local processing (near sensor), has been advocated for e-skin. Towards this end, we present the printed silicon nanoribbons (Si NRs) based multi-gate field-effect transistors (v-FETs) as the building blocks. The direct roll transfer printing (DRTP) is employed to integrate well-defined Si NRs arrays on flexible substrate, followed by conventional microfabrication steps to deposit the dielectric and metal electrodes needed for fabrication of v-FETs. The capacitively coupled top-gates and the floating gate (FG) of v-FETs allows mimicking of neural functions. The transfer characteristics, obtained by biasing the two top-gates, showed distinct electrical responses defined by the top-gate electrode width. The top-gate (G1) with a higher width (weight), shows stronger gate coupling, two order of higher current on/off ratio when compared with the top-gate (G2) having a lower width (weight). The threshold voltage of these flexible v-FETs could be tuned/adjusted to implement learning process, thereby offering an attractive opportunity to build spiking neurons and circuits for e-skin with neural-like computing capability.