Highly sensitive stretchable sensor combined with low-power memristor for demonstration of artificial mechanoreceptor properties

Abstract

The development of high-performance sensors emulating the human skin response to external mechanical stimuli is of increasing importance in the era of artificial intelligence and robotics. Towards the realization of artificial skin, various parameters must be met including flexibility, biocompatibility and low-power consumption of the employed sensors. In parallel, a multisensory platform for both the detection and storage is required for emulating the physical properties of human skin. Under this perspective, in this work, an extremely sensitive resistive stretchable sensor that can achieve a gauge factor (GF) of ~107 was firstly demonstrated based on the employment of a polydimethylsiloxane (PDMS) substrate decorated with Pt nanoparticles (NPs) as the stretch-sensitive medium placed in between of two Ag electrodes. A critical step to achieve such performance is the formation of a rippled surface of the PDMS substrate through the combined use of pre-stretch and the deposition of a thin Al2O3 film by atomic layer deposition (ALD) technique that enables the fabrication of highly stretchable Ag electrodes. The wavelength of the ripples, as well as the peak-to-valley height among them, can be directly controlled by tuning the applied pre-stretch load on the PDMS. By taking advantage of the extreme sensor sensitivity achieved, the emulation of the functionality of a biological mechanoreceptor was further demonstrated by connecting in a parallel circuit configuration the sensor with a SiO2-based conductive-bridge memory. Various synaptic properties of the above sensory system were demonstrated, including the paired-pulse facilitation (PPF) and long-term plasticity (LTP), indicating the capabilities of our system to perform neuromorphic computations at the edge.