Gabriele Boschetto1,Stefania Carapezzi1,Aida Todri-Sanial1,2
LIRMM, University of Montpellier, CNRS1,Eindhoven University of Technology2
Gabriele Boschetto1,Stefania Carapezzi1,Aida Todri-Sanial1,2
LIRMM, University of Montpellier, CNRS1,Eindhoven University of Technology2
The Internet of Things (IoT) is projected to become an hyper-linked system with 30.9 billion of connected devices by 2025. A game-changer step for IoT further evolution will be represented by the development of smart objects that are context- and self-aware, with ability to “self-assemble” in smart units. In this respect, IoT sensory nodes are the key elements to embed ambient-awareness in IoT. Currently, IoT sensory nodes are an ensemble of four basic units, each one dedicated to a specific role: one for sensing, one for processing, the other two for transceiving signals and power supply. The stimuli incoming from the ambient are the inputs of the sensing module, whose output are unorganized data to be fed into processing module. The physical separation between sensing and processing units is the bottleneck for the performance of sensory systems, especially given the present deluge of sensory data that need to be analyzed. Thus, novel approaches are required to enhance the efficiency of IoT sensory nodes.<br/>Recently, the in-sensor computing approach has gained the spot-light from both academic and industrial research, which consists into realizing the integration of sensing and computing functions into a single element. Indeed, this corresponds to mimic the elements of nervous system dedicated to sensory transduction of environmental information, the sensory neurons. In sensory neuron devices the response characteristics becomes not linear, and the thorough understanding of the physical mechanisms which connect the ambient input to the device output is key to control their response characteristics. In the present work, we explore the implementation of a sensory neuron device based on vanadium dioxide (VO<sub>2</sub>) volatile memristor. Since the temperature triggers a volatile resistive switching in VO<sub>2</sub>, volatile memristors can be fabricated as two-terminal VO<sub>2</sub> devices, where Joule effect is responsible for the resistance change of VO<sub>2</sub> channel. It is possible to fabricate compact and scalable relaxation oscillators [1] by inserting VO<sub>2</sub> volatile memristors into a simple RC circuits, which model the oscillations behavior of biologic neurons. We avail of a dedicated technology computer-aided design (TCAD) approach [2 - 4] to perform 3D electrothermal simulations of VO<sub>2</sub> neuron oscillators. We simulate the modulation behavior of the external temperature over the output of such devices, and thus we unravel the physical mechanism behind the temperature sensing of VO<sub>2</sub> neuron device. The insight we gain from our TCAD simulation is primordial to fully exploit the VO<sub>2</sub> sensory neurons.<br/><b>Acknowledgments.</b> Authors wish to thank Dr. S. Karg, IBM Research Europe, Zurich, Switzerland, for providing the experimental data used for the calibration of the TCAD model and the useful discussions about the experimental devices. Authors also wish to thank Dr. A. Nejim and Dr. A. Plews, of Silvaco Europe Ltd., Cambridgeshire, United Kingdom, for providing the customized version of PCM model [2] used to simulate the VO<sub>2</sub> material as well as for the useful discussions about the TCAD and mixed mode simulations.<br/>[1] E. Corti et al., Front. Neurosci., vol. 15, 2021. DOI: 10.3389/fnins.2021.628254<br/>[2] ”Victory Device User Manual”, version 1.19.1.C, Silvaco Inc<br/>[3] S. Carapezzi et al., IEEE J. Emerg. Sel. Topics Circuits Syst. 11, 4, 2021. DOI: 10.1109/JETCAS.2021.3128756.<br/>[4] S. Carapezzi, et al., MRS Communications, 12, 427–433, 202. DOI: 10.1557/s43579-022-00196-3