A “Smart” Gas Sensing System Composed of Micro-Hotplates and Artificial Neural Network

Abstract

Integrating in one package a gas sensor array (GSA) unit and an artificial neural network (ANN) unit, a gas-sensing system based on two types of metal-oxide semiconductors is described. The GSA consists of micro-hotplates (MHPs) monolithically integrated with thin-film transistors (TFTs) built using a first metal-oxide semiconductor (MOS). The structural plates for realizing the MHPs are spontaneously suspended over a sealed cavity without requiring a sacrificial layer etch, thus largely overcoming potential issues of process incompatibility between the construction of the MHPs and the TFT active-matrix circuits. A second gas-sensing MOS is coated on the MHPs to reduce process complexity, with the resulting low sensing specificity remedied by exploiting the different temperature-dependent sensitivities of the different gas species. ANN consisting of dual-gate TFTs has been fabricated and trained for inferencing gas identification and quantification. This technology can be more generally applied to realizing other micro-fabricated “smart” sensors. [2023-0159]