Developing Gallium Nitride Nanostructured Sensors for Resolving Cross-sensitivity to Gases
Date
2020
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
There is a great need for the development of highly selective sensors for detecting various toxic gases and their mixtures in many industrial, medical, space exploration and environmental monitoring applications. Environmental gases such as SO2, NO2, ethanol and H2 are harmful either to the environment and/or to living beings and their monitoringrequires sensors capable of detecting ppm level of these gases well below their Occupational Safety and Health Act (OSHA) permissible exposure limits. Metal oxide based sensors to detect these environmental pollutants have been the subject of intensive research for several decades. However, these metal oxide sensors lack precise selectivity towards any specific gas.
In this work, GaN nanowires have been formed on a Si substrate using production standard stepper lithography and top-down approach. Different functionalized deviceswere prepared by the deposition of metal oxides- TiO2, ZnO, WO3 and SnO2 by optimized RF sputtering on nanowires followed by rapid thermal annealing. The elemental composition, crystallinity, and surface topography of metal-oxide/GaN nanowires were fully characterized. The gas sensing data was collected and analyzed for all four sensors. To examine the real-world applicability of the fabricated sensor devices,
their additional sensing properties, including gas sensing adsorption and desorption rate, cross-sensitivity to interfering gases, and long-term stability at various environmental conditions were investigated.
Fundamental electronic interactions and thermodynamics between the gas molecular adsorption on ideal metal-oxide surfaces have been investigated with Density Functional Theory (DFT) molecular simulations. The functionalized GaN in contact with gas molecule was designed and geometrically optimized. Simulation results revealed that TiO2 and ZnO functionalization enabled the most energy favorable surface for NO2 and SO2 adsorption, respectively. In addition, the electronic properties of these oxide functionalized GaN have been studied in terms of the total density of states (TDOS) and projected density of states (PDOS), indicating an excellent agreement with the abovementioned experimental measurements.
A gas sensor array has been designed and developed comprising of functionalized GaN nanowires using industry standard top-down fabrication approach. The receptor metal/metal-oxide combinations within the array have been determined from prior molecular simulation results. The gas sensing data was collected for both singular and
mixture of gases under UV light at room temperature. Each gas produced a unique response pattern across the sensors within the array by which precise identification of cross-sensitive gases is possible. Unsupervised principal component analysis (PCA) technique was applied on the array dataset. It is found that each analyte gas forms a separate cluster in the score plot for all the target gases and their mixtures, indicating a clear discrimination among them. Then, supervised machine learning algorithms such as Decision Tree, Support Vector Machine (SVM), Naive Bayes (kernel), k-Nearest
Neighbor (k-NN), and artificial neural network (ANN) were trained and optimized using their significant parameters for the classification of gas type. Results indicate that optimized SVM and NB classifier models exhibited 100% classification accuracy on the test dataset. Statistical and computational complexity results indicate that back propagation neural network stands out as the optimal classifier among the considered ANN algorithms. Then, ppm concentrations of the identified gases have been estimated using the optimal model. Furthermore, implementation of the developed sensor array in combination with neural network algorithm for real-time gas monitoring applications has been discussed.
In another work, TiO2 functionalized GaN nanowire-based back-gate FET device has been designed and implemented to address the well-known cross-sensitive nature of metal oxides. Even though a two terminal TiO2/GaN chemiresistor is highly sensitive to NO2, it suffers from lack of selectivity toward NO2 and SO2. Here, Si back-gate with CAlGaN as the gate-dielectric has been demonstrated as a tunable parameter, which enhances discrimination of these cross-sensitive gases at room temperature (20 ᵒC). Compared to no bias, back-gate bias resulted in a significant 60% increase in NO2
response, whereas the increase is an insignificant 10% in SO2 response. Sensor die/process and packaging reliabilities of metal-oxide/GaN nanowire-based gas sensors have been studied for the first time, using industry standard accelerated lifetime tests, such as- High Temperature Operating Life, High Temperature Storage Life, Temperature Cycling Test and Highly Accelerated Stress Test. The metal-oxide functionalization used for sensing ethanol exposure in this study is ZnO. For all the tests, sample ZnO/GaN devices have been exposed to 500 ppm of ethanol in dry air at room temperature (20 ℃) to observe and record the degradation of signal to noise ratio (SNR) as a function of stress time and number of thermal cycles. Although no complete device failure was observed in any of the performed tests, gas sensing response kept decreasing gradually due to increasing stress. The lowering of the sensor response is believed to be due to gradual phase transformation of the receptor ZnO and baseline resistance increase. The method for estimating failure rate and lifetime of sensor devices has been developed.
Using statistical data from the performed accelerated stress tests, chi-square distribution has been implemented to predict the failure rate and lifetime of GaN nanostructured sensor devices. Total failure rate combining die and package failure rates was found as 7.09 x10-4 day-1. The mean-time-to-failure (MTTF) of the stressed devices was estimated
about 4 years, representing the lifetime of the GaN nanostructured sensors.