Batch nanofabrication of nanoheater for power-efficient gas sensors and wireless IoT application

Abstract

The increasing need for affordable and low-power gas sensors is driven by the rapid growth of wireless applications in various monitoring systems, such as stationary, drone-based, and wearable gas monitoring devices. Recently, their use has expanded to include areas like home indoor air quality monitoring, early disease diagnosis through breath analysis, and food spoilage detection. Among the different sensing materials, metal-oxide semiconductors (MOS) are often chosen for gas sensors in wireless sensor networks due to their cost-effective production, ease of design, high sensitivity, and suitability for making small sensors. MOS gas sensors work by detecting the target gas through changes in electrical resistance when the gas is present. However, inherent issues of MOS-based gas sensors such as high power consumption due to heater-based operation and limited selectivity have posed obstacles to their widespread internet of things (IoTs) application. Electronic nose (e-nose) systems, employing a gas sensor array to enhance selectivity for accurate gas identification, paradoxically lead to increased power consumption and costs proportional to the number of sensors used. The challenge lies in balancing the quest for precise gas identification—with the desired high selectivity—against efficiency concerns, including power use, size, and expense. This dissertation presents a method to produce one-dimensional (1D) nanoheater by coating a suspended pyrolyzed carbon nanowire with a thin layer of gold, making it a resistive heater, and using it in a portable gas sensor system. This advanced nanostructure is made without complex nanofabrication and alignment processes, thanks to its suspended design and built-in shadow mask. The carbon nanowires are produced in batches using carbon- microelectromechanical systems (C-MEMS) technology and remain strong and functional during further nanopatterning due to their excellent mechanical robustness. The nanoheater is used in gas sensors by precisely placing metal oxide semiconductor nanomaterials onto the center of the nanoheater, ensuring uniform heating and reliable gas detection. The nanoheater is integrated into an ultra-low-power MOS sensor-based e-nose system for real-time gas identification, addressing high power consumption and limited selectivity issues in conventional MOS sensors. This e-nose system employs a single MOS sensor built on a suspended 1D nanoheater, driven by duty cycling. Duty-cycling is an energy-saving method of electrothermal Joule heater by periodically turning on and off the heater power. This method reduces the active time of devices (i.e., duty cycle), resulting in a significant reduction in average power consumption. This strategy also induces distinct gas sensing behaviors at high and room temperatures, enabling a single sensor to provide two independent and coupled outcomes. The sensor's ultrafast thermal response, enabled by its miniaturized sensor’s size, effectively decouples the effects of temperature and surface charge exchange on the MOS nanomaterial’s conductivity. As a result, two distinct sensing signals, alternating between responses coupled with and decoupled from thermally enhanced conductivity. The magnitude and ratio of these dual responses vary depending on the gas type and concentration, facilitating the early-stage gas identification of five gas types within 30 seconds via a deep learning algorithm. Compared to other pioneering machine-learning-based e-nose systems, our approach demonstrates remarkable performance. Despite utilizing a heater-embedded MOS sensor (sensor temperature ~250 °C), our system achieved the lowest power consumption at just 160 μW, which is 99% lower than that of microheater-based systems and 65% lower than micro- LED-based systems. Leveraging a convolutional neural network (CNN) enabled real-time prediction capabilities, with type classification exceeding 90% accuracy and a regression error for concentration prediction under 20%. Therefore, this approach addresses the critical trade- off between minimizing power usage and maintaining high selectivity. Additionally, all fabrication steps for the suspended 1D nanoheater-based MOS sensors were based on wafer-level batch microfabrication processes, despite the sensor’s complex 3D mixed-scale architecture, ensuring cost-effective manufacturing. This achievement is notably superior, reflecting our system's efficiency and effectiveness in gas identification, and meets efficiency concerns including power use, size, and expense.