Search papers, labs, and topics across Lattice.
This study explores the concept of using the physical properties of conducting polymer dendrites to store sensory data in a neuro-inspired electronic nose, effectively mimicking natural systems' evolutionary processes. By exposing sensing elements to volatile molecules, the impedance of the interconnects changes, enabling the electrochemical growth of dendrites that serve as a form of "passive memory." The findings suggest that this approach can significantly reduce manufacturing complexity and enhance the calibration of dense sensing arrays for specific environmental exposures.
Memory in electronics can now evolve like in nature, using low-resource and low-energy methods to store sensory data in physical structures.
If electronics drives only electrons to charge electrodes, natural systems learn by moving matter to evolve. Morphogenesis in sessile organisms can be seen both as a fabrication process and as an operative mechanism. However, intricating manufacturing and programming functionalities in electronic hardware is not conventional. In this study, we experimentally implement such a concept of an evolutionary electrical system using a neuro-inspired electronic nose as a model, to store a history of sensory data in the physical properties of the electrical interconnects of sensing elements. Triggered only by volatile molecule exposures, different sensing elements change instantaneously and reversibly their impedance, so pulse voltages enable the electrochemical growth of conducting polymer dendrites. The strength of the evolving interconnects is specific to the sensing materials and to the nature of volatile molecules to which they are exposed. The dendritic growths occur exclusively when exposed to volatile samples, and stop immediately after interrupting the exposure. The capability of such"passive memory"was also assessed by simulating a network architecture, which showed that this way of storing information should greatly diminish the fabrication complexity of a highly dense sensing array while realistically enabling its calibration to classify user-specific environment exposures. By demonstrating that memory in electronics can be a concept linked to manufacturing like in living organisms, this study shows that low material resources and low energy activation can be exploited for practical electronic applications in future-emerging sensing technologies.
