Synthesis and characterisation of Cu 0.5 Mg 0.5 Fe 2 O 4 ${\text{Cu}}_{\mathbf{0.5}}{\text{Mg}}_{\mathbf{0.5}}{\text{Fe}}_{\mathbf{2}}{\mathbf{O}}_{\mathbf{4}}$ nanoparticles doped with cadmium by co-precipitation method for acetonitrile, acetone, and ethanol gas detection with deep learning-based methods

In this study, a magnetic disk was prepared using nanoparticles with a diameter of less than 15 nm. The morphological and structural characteristics of these nanoparticles were systematically examined using X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and alternating force gradient magnetometry (AGFM). XRD analysis confirmed that the average diameter of the copper–magnesium ferrite nanoparticles doped with cadmium was approximately 12 nm, consistent with TEM results, which also showed uniform particle distribution and a tendency to form clusters in powdered form. AGFM measurements revealed that the magnetic property of the powder sample was 15.83 emu/g, which increased to 22.70 emu/g after compression, highlighting the influence of particle density and morphology on magnetic behaviour. Gas sensing tests demonstrated that the fabricated sensors achieved exceptional sensitivity, particularly to acetonitrile, with a maximum sensitivity of 92.3%. A hybrid deep learning model, Bi-LSTM, was utilised to enhance the precision of gas classification. The proposed methodology was benchmarked against traditional machine learning models, including LSTM and RNN, and demonstrated superior performance. The accuracy of gas detection reached an impressive 99.89%, as validated by ROC analysis, underscoring the efficacy of the deep learning-based approach. These findings highlight the potential of cadmium-doped ferrite nanoparticles for high-performance gas sensing applications, suitable for both industrial and medical uses.