In-Sensor Computing by Soft Threshold Logic Gates Under Different Humidity Conditions

Abstract

In pursuing machine intelligence comparable to biological systems, the crucial role of the body in enabling efficient data processing is often overlooked. In contrast to silicon-based architectures, biological organisms exploit the intrinsic properties of their materials and morphology to support sensing, actuation, and low-energy information processing. This perspective has motivated the development of embodied intelligent systems, where computation emerges from the physical interactions between materials and the environment. Drawing inspiration from these principles, artificial systems can address the challenge to integrate sensing, actuation, and control systems directly within their materials, improving processing efficiency and autonomy, in an unconventional, albeit biologically-inspired strategy. Here we demonstrate how soft nanocomposite materials, based on gold cluster-assembled thin films implanted in polydimethylsiloxane (PDMS) substrate, can perform reliable processing in ambient environmental conditions. We exploit the nonlinear electrical behavior of this material to implement soft reconfigurable nonlinear threshold logic gates able to work in different ambient conditions. In particular, this work investigates how humidity influences the resistive switching and computational capabilities of the nanocomposites, with the goal of developing multifunctional materials that combine their sensing capabilities to environment and mechanical stimuli with computation on a soft substrate able to support adaptive, energy-efficient embodied intelligence for edge computing devices.