Detection of mustard oil adulteration by VOC fingerprinting using a CNT-based electronic nose coupled with LC-MS profiling and supervised learning.

Mustard oil possesses significant nutritional and medicinal value, and its adulteration with cheaper oils poses serious health and economic concerns. This study presents a qualitative, non-destructive approach for detecting mustard oil adulteration using a carbon nanotube (CNT)-based electronic nose (E-nose) integrated with liquid chromatography-mass spectrometry (LC-MS) based molecular profiling and supervised machine learning. Volatile organic compound (VOC) profiles of six cold-pressed edible oils (mustard, canola, cottonseed, palm, rice bran and soybean) were characterized using untargeted LC-MS confirming distinct VOC fingerprints. Binary mixtures of adulterated mustard oils (1-80%) were subsequently analyzed using the E-nose. Sensor-derived VOC features were evaluated using unsupervised (PCA, t-SNE) and supervised (LR, LDA, SVM) models. Among the classifiers, SVM achieved the best overall performance with mean classification accuracies of 98.63-99.19% across different adulteration systems. The proposed CNT-based E-nose platform demonstrates strong potential as a rapid, portable and cost-effective screening tool for edible oil authentication and quality assurance.