Application of AlGaN/GaN Heterostructures for Ultra-Low Power, Low Noise Nitrogen Dioxide Detection

As health care costs rise, there has been resurgence in preventive care and personal monitoring. While the effects of toxic gases such as, nitrogen dioxide (NO₂), carbon monoxide (CO) and other toxic gases have been studied in static environments, current sensors cannot be used to monitor personal dynamic exposure to these gases. Toxic gas concentrations can vary greatly within small areas due to industrialized sectors and vehicle traffic. In the case of NO₂, it has been associated with decreased lung function and increased asthma and respiratory infection [1]. This has created the need for low-power, portable, and non-invasive monitoring devices, which may be fulfilled by Gallium Nitride (GaN) heterojunction devices that reversibly react to surface adsorption.

GaN heterojunctions have a 2 dimensional electron gas (2DEG) plane below the AlGaN layer that is able to be modulated by surface state changes. [2]. This effect is due to piezoelectric charge mirroring from surface states down to the channel. Gas detection on GaN is possible because the surface is chemically stable and there are naturally occurring adsorption sites (i.e. surface states) allowing for fast, reversible, low-noise detection by use of the high mobility 2DEG resistance changes. This work investigates gas response effects to varying exposure levels of nitrogen dioxide (NO₂) and optimization of sensing parameters for low-power low-noise exposure tracking. Sensitivity to various concentrations of NO₂ were measured between 50ppb - 500ppb. This work focuses on low-power operation, thus room temperature is chosen for detection with the option of heating to perform sensor recovery.

Two-terminal AlGaN/GaN devices were fabricated by PVD of Ti/Al/Ni/Ti/Au followed by an ICP-RIE using BCl₃ to isolate the 2DEG between devices. The devices were then annealed at 850°C for 30s to obtain an Ohmic contact to the 2DEG. The surface was then functionalized by Atomic Layer Deposition (ALD) with 6nm of SnO₂. SnO₂ is a well-known material for gas detection and has been utilized for oxygen vacancy adsorption sites, while maintaining a low-noise signal from the 2DEG.

Testing was performed in a custom gas testing chamber equipped with a heating chuck and commercial sensors for gas concentration calibration. The device bias voltage was set to V_B=50mV and NO₂ was flown into the chamber at room temperature for 5 minutes, followed by a 5 minute rest period. During each exposure, device current was measured and resistance was extracted. Fig. 1 demonstrates the sensor response to repeated 500ppb exposures. With increasing exposure count, the sensor’s increase in resistance change and slope decreases. The device saturates during the rest period, without demonstrating recovery due to the lack of thermal energy to desorb NO₂ from the surface sites. Interestingly, the rate of change in resistance (DR/Dt) plateaus during the exposure. By characterizing change in resistance and rate of change in resistance, it is possible to monitor the dose of NO₂ before device reset. Fig. 2 demonstrates the sensor response to varying levels of NO₂. The magnitude of slope and resistance change increases with each increase in NO₂ concentration, showing good sensitivity to various levels of exposure. As the device is exposed to NO₂, the channel resistance increases and the sensing power reduces for greater concentration values. Larger response could be seen with a heated substrate, but has been limited in this work to reduce total power consumption and for device surface recovery, as seen in Fig. 3. Since the device is operated at room temperature, the peak operational power is during rest periods. This power was calculated to be approximately 1.25µW for the devices tested and should allow for long-term operation and wearability.

Using AlGaN/GaN heterojunctions, ultra-low power, low-noise detection of NO₂ was demonstrated at room temperature. The detection concentrations varied from 50ppb to 500ppb and showed distinguishable difference in response to each concentration. The maximum power recorded was 1.25µW and the devices show recovery (i.e. NO₂ deadsorption) with substrate heating as low as 100°C.

This work was supported by ASSIST ERC Program of the National Science Foundation under Award Number EEC-1160483.