Research On The Rapid Adulteration Detection Of Edible Oil By Ion Mobility Spectrometry

Edible vegetable oils plays an important and indispensable role in daily diet, providing energy, fatty acids, vitamins, polyphenols for human being, which possess highly nutritional value. Therefore, the quality of edible oils is closely related to the human health. Nowadays, the adulteration detection methods of edible oils were mainly based on large instruments, which are time-consuming, complex sample pretreatment and high test cost, and they were unsuitable for site testing. According to the characteristics of decentralized production and business operations in China, a simple, fast, portable detection technology(ion mobility spectrometry) was applied to the quality analysis for edible oils in this thesis. Moreover, chemometrics methods were combined to establish the adulteration detection methods of edible oils.1. The database of IMS fingerprint spectra of main edible vegetable oils was builtAfter optimization for detection parameters, the optimized experimental conditions were determined as follows: the inlet temperature 170 ℃, the temperature of drift tube 60 ℃, the analysis time 20 s. Each edible oil sample diluted in n-hexane by 50 times was injected with 4.0 μL. The IMS data at the drift time of 11.73-28.6 ms were selected to eliminate the influence from solvent. The spectra at detection time between 10.33 and 10.67 s were summarized and normalized to the maximal value to create the standard IMS fingerprint spectra of this oil sample. Representative oil samples from the main producing areas in China were collected and employed to build the database of IMS fingerprint spectra of edible vegetable oils, including rapeseed oils, peanut oils, sunflower seed oils, soybean oils, sesame oils, flaxseed oils, grape seed oils, walnut oils and cottonseed oils. Taking grape seed oils, flaxseed oils, rapeseed oils, sesame oils and soybean oils for example, the specificity of IMS fingerprint spectra for different kinds of edible oils was investigated. As results, PCA score plot showed good clustering among different edible oils. Furthermore, 3 fold cross-validation results of random forest discriminant model showed 5 kinds of oils could be completely correctly distinguished, indicating the IMS detection technology and the database of IMS fingerprint spectra of edible vegetable oils could provide an effective data and technical support for adulteration detection of edible oils.2. Detection method of fried oils was built based on IMSThe IMS spectra of 27 deep-fried oil samples and 28 edible oil samples were analyzed by PCA and Recursive support vector machine(R-SVM) after data pretreatment by log-transformation and Pareto scaling. PCA three-dimensional score plot showed the two kinds of oils were completely separated and there was no overlap in the space. The result of 10 fold cross-validation of model built by R-SVM also showed highly correct recognition rate for deep-fried oils, which was at 98.8%. Therefore, IMS could be a new fast detection method for deep-fried oils with high accuracy.3. Adulteration detection methods of edible oils were established by IMSTaking flaxseed oils, sesame oils and grape seed oils for examples, which possess high risks of being adulterated with other cheap edible oils or even being imitated by low-price oils with essences. After simulation of adulteration of those three edible oils, R-SVM and RF were applied for data analysis. As a result, the RF classification model could better classify three types of edible oils with their adulterated oils, respectively. The detection results indicated that the discriminant model built by R-SVM could identify adulterated edible oil samples(≥5%) with high accuracy above 90%, for the adulterated sesame oil samples(≥10%), the classification rate is highest at 94.2%. The PCA result for adulteration of sesame oil essence showed the IMS spectra of counterfeit sesame oils were completely separated from pure sesame oils, and two types of counterfeit sesame oils also had a separation tendency, as displayed on the 1st and 2nd PCs. The discriminant model established by Random Forests also proved the pure sesame oils and two types of counterfeit sesame oils could be completely separated, indicating IMS spectra could not only recognize the counterfeit sesame oils, but also clearly identify the cheaper vegetable oils used to prepare the counterfeit sesame oils.