Dissertation > Industrial Technology > Light industry,handicrafts > Tobacco Industry > Product standards and testing

Study on the Determination of Trace Metal Elements in Tobaccos and Pattern Recognition

Author SongShiQiu
Tutor SunYuAn
School Zhengzhou University of Light Industry
Course Applied Chemistry
Keywords tobacco race element inductively coupled plasma mass spectrometry cluster analysis support vector machine
Type Master's thesis
Year 2010
Downloads 74
Quotes 0
Download Dissertation

Tobaccos need kinds of trace elements for its life activities, and they influence the quality and yield of the tobaccos corporately. But tobacco plants very easily take up some toxic elements and concentrate them in leaves, specially some heavy metal elements which are harmful to human health. There is a deep relationship between the contents of trace elements in tobaccos and quality of tobaccos, and the relationship between the contents of trace elements in tobaccos and human health is also indispensable. Determination of trace metal elements in tobaccos provide the information of trace elements contents for cultivation of tobaccos and smoking and health and a scientific basis for quality control of tobaccos.The trace elements contents differ in different kinds of tobaccos due to the different geographical origin and the way of cultivation, and therefore the contents of trace elements can be very useful as markers for identification of the product’s geographical origin and authenticity. Classification of tobaccos based on their contents of trace elements is helpful for identification of tobacco classes and geographical origin. The main contents of this dissertation are as followings:1. The influence of contents of trace elements on tobaccos quality and human health was reviewed. The popular analysis techniques of determation of trace elements in plant were introudced, specially a detailed introduction of the applation of inductively coupled plasma mass spectrometry technique was gave. Pattern recognition technique as a normal classfication technique for plant samples was introduced and the application of clustering analysis and SVM classification was mainly summarized.2. The method of tobaccos sample pretreatment was improved. The result of eight elements was used as examination index. There was Ti, Ni, Cu, Rb, Sr, Y, Sb, Tl. 2 mL 50% HNO3 and 1 mL 47% H2O2 were chosen for tobaccos digestion, and the acidity of final digestion solution was confined to about 2%. Standard solutions were diluted with 2% HNO3. The digestion solutions were determined by ICP - MS directly after adding internal standard elements Ge and Rh. Compared with other digestion menthods, this method not only can save the process of drying or dilution and extended the life of relevant components of the instrument, but also the errors of the inconsistency between digestion solutions and standard solutions could be reduced. The detection limits were 0.006 - 1.133 ng/mL. The relative standard deviations were 0.90% - 5.66%. The recovery rates were 93% - 102%. The results showed that this method proposed could be applied to multielement simultaneous determination of tobaccos. The improved method was applied to determine the contents of 44 trace elements in 88 tobaccos from six areas. The contents of 44 trace elements in 88 tobaccos sample were got, and the results can provide more comprehensive informations of trace elements for cultivation of tobaccos and smoking and health. From the results, we found there had certain relevance between the contents of trace elements in tobaccos and the geographical origin.3. Fuzzy c - means (FCM) was used for classification of the different kinds of tobaccos based on the contents of trace elements. The results show that there was a good correspondence between the results of FCM clustering analysis and the geographical origin of the tobaccos, when the samples from Shandong Provice and the contents of the rare earth elements were removed. The accuracy of Henan and Sichuan area reached 100%. The accuracy of Fujian and Yunnan area were 88.89% and 80.00% respectively. But the accuracy was low when all the testing results including rare earth elements were employed for FCM clustering analysis. It indicated that selection of characteristic elements was important for successful classification when FCM clustering analysis was used for tobaccos classification.4. Support vector machine (SVM) classification was introduced for classification and prediction of the tobaccos sample from different origin. The contents of 44 trace elements were input features and the corresponding output were the diffrerent origin of the tobaccos. The results showed that the accuracy of the prediction reached 90%, when all the tobaccos sample and trace elements were involved. The accuracy of the prediction could be improved to 92.59%, when the influence of uniformity of the samples and the contents of the rare earth elements were considered. The results demonstrated that the SVM classifier has an excellent performance, high accuracy of the prediction, and powerful generalization capability. SVM classification was superior to FCM analysis, and it could be applied to classification and prediction of tobaccos.

Related Dissertations
More Dissertations