Monolithic photoactivated gas sensor based on micro light-emitting diodes and functional nanomaterials

Abstract

In this study, monolithic photoactive gas sensors based on semiconductor metal oxide (SMO) nanomaterials directly integrated on gallium nitride (GaN) micro light-emitting diodes (μLEDs) were developed to maximize the energy transfer efficiency. Compared with the conventional thermal-activated and photoactivated gas sensors, it could achieve comparable sensing performances, significantly reduced power consumption (microwatt-level), and outstanding mechanical reliability. The proposed μLED platforms were fabricated through well-established micro-fabrication processes, and SMO nanomaterials (i.e., zinc oxide (ZnO) nanowire and porous indium oxide (In2O3) thin film), which have a large surface-to-volume ratio, were integrated through the hydrothermal reaction or the glancing angle deposition method. First of all, monolithic photoactivated gas sensors, which are composed of gas-sensitive ZnO nanowires (Egap = 3.2 eV) and ultraviolet μLEDs (λpeak = 390 nm), were developed. Current densities, irradiances, and external quantum efficiencies (EQE) of various sized-LED platforms (emission area = 30 × 30 to 200 × 200 μm2) were comprehensively evaluated. The μLEDs showed improved irradiance and energy conversion efficiency as the size of LEDs was reduced from 200 × 200 μm2 (EQE of 4 %) to 30 × 30 μm2 (EQE of 9 %). Furthermore, the NO2 sensing performance (i.e., sensitivity) at optimal operating power was improved by the miniaturization of LEDs because of the improved emitting uniformity. In specific, the most miniature gas sensor (active area = 30 × 30 μm2) showed excellent NO2 sensitivity (ΔR / R0 = 605 % to 1 ppm NO2) at the optimal operating power (~184 μW). Second, blue μLED platforms (λpeak = 430 nm), which are well known to have the highest energy conversion efficiency in GaN-based LEDs, were newly developed. Porous In2O3 films (Eg = 3.6 eV) were integrated on the blue μLED platform, and silver (Ag) nanoparticles were coated on the surface of In2O3 to increase the absorbance to the blue-range light. Ag nanoparticles absorb the blue light by the plasmonic effect and spontaneously transfer excited electrons to the surface of In2O3. Consequently, the EQE of the μLED platform, the optimal power consumption, and the response (ΔR / R0 (%)) to 1 ppm NO2 were ~ 17.3 %, ~1 μW, and ~1319%, respectively. Lastly, a novel operation strategy, i.e., time-variant illumination of the monolithic photoactivated gas sensor, is introduced for selective identification of various gas species. The excellent mechanical stability of the μLED platform enables this-like operation. The difference in chemical kinetics of variant gas reactions induces the variety in the forced transient sensor signal under the time-variant illumination. Here, a deep-learning-based real-time classification algorithm was introduced to analyze frequency spectra from a single sensor. Therefore, the proposed method achieved high classification (~99.96 %) and quantification (91.1 %) accuracies for various toxic gases (carbon monoxide, ethanol, acetone, and methane). The developed ultra-low-power gas sensor is highly expected to realize various applications such as personal environmental monitoring devices, smart factories, smart farms, and personalized healthcare services by combining with battery-driven internet of things (IoT) mobile platforms.