Ultra-high ammonia gas response of phase-stabilized (Fe0.2Ni0.2Cr0.2Mn0.2Zn0.2)3O4-δ high-entropy spinel oxide sensor array and its machine learning predictions

In this work, the gas sensing performance of phase-stabilized (FeNiMnZnCr)₃O₄ high-entropy spinel oxide (HSO) gas-sensors via screen-printing were investigated, where the HSO powders were synthesized via solution combustion synthesis (SCS) using three different fuels: citric acid, urea, and glucose. Although all HSO powders were obtained at 500 °C, the formation of stable spinel phase was evidenced at 600 °C. Among all fabricated sensors, G-800 gas sensor depicted a stable ultra-high response of ∼3471 towards 100 ppm of ammonia gas along with a notable response of ∼162 even at 10 ppm (where G means glucose and 800 represents calcination temperature in °C) and it demonstrated a strong device-to-device reproducibility with stability of ∼35 days. A synergy of crystallinity and increased porosities from XRD and FESEM micrographs resulted in ultra-high gas-response towards ammonia gas compared to volatile organic compounds such as formaldehyde, methanol, and ethanol). The presence of defect band and oxygen vacancies observed from the Raman and XPS analysis, were complemented by the presence of porosities confirmed from BET surface area analysis. Subsequently, the machine learning (ML) algorithms are applied on sensor signals to estimate the concentration of ammonia gas, and among all the ML classifiers, RFC gave reasonably better predictions in three concentrations regimes with a good classification accuracy of 93.3 ± 5.3 %, 90 ± 7.5 %, and 83.3 ± 13.1 % for G-600, G-700, and G-800, respectively. The proposed ML studies enable accurate detection of hazardous ammonia levels using HSO-based sensors, showing strong potential for integration into diagnostic platforms targeting ammonia breath markers.