Functionalized Metal Oxide Semiconductor Nanofibers Based Electronic-nose (E-nose)

Low-cost portable or wearable chemical sensing systems with high sensitivity and selectivity can significantly impact our daily life by providing real-time information to improve the health and safety of humans. Herein, 91 different tungsten trioxides (WO₃) and tin oxide (SnO₂) polycrystalline nanofibers with different dopants (e.g., Au, Ag, Pd, Pt, Ru, Sr, Ni, Cu, Co), diameter (i.e., 23 to 300 nm), crystal structure, and crystallinity (i.e., nano to micro-crystalline) were systemically synthesized by electrospinning, characterized, and correlated to sensing performance.

In Chapter 1, the latest developments in the one-dimensional metal oxide semiconductors (MOS) based chemiresistive gas sensors are critically reviewed.

In Chapter 2, one part per billion (ppb) level hydrogen sulfide detection was demonstrated by optimizing the grain size of gold and tungsten trioxides.

In Chapter 3, selective detection of methyl salicylate, a chemical warfare agent simulant, was achieved from platinum and gold decorated tungsten trioxides. Enhanced gas sensing performance was correlated with quantitatively determined energy barrier and oxygen ionic species.

In Chapter 4, portable electronic-nose (E-nose) consisting of 16 temperature tunable MEMS sensor array with controlled electronics were prototyped and utilized to collect train data set toward 14 different inorganic (i.e., ammonia, hydrogen sulfide, carbon monoxide, nitric oxide, nitrous oxide) and volatile organic compounds gasses (i.e., benzene, toluene, ethyl benzene, p-xylene, methyl salicylate, acetone, ethanol, methane) at different operating temperature (300 to 450°C).Different machine learning models (e.g., Random Forest (RF), K-Nearest Neighbor (KNN), and Neutral Network (NN)) were trained and tested for sensing features down selection and E-nose performance evaluation. Using RF trained algorithms, our portable E-nose system consisted of 16 different metal oxide nanofibers and was able to identify 14 analytes and 98 concentrations with accuracies greater than 95%.