The Correlation Between LF-NMR Relaxation Propertiesand Fatty Acids Compositionof Sunflower Oilduiring Deepoxidation Processes And Its Applycation

This paper was supported by the The National Natural Science Foundation of China(NSFC31201365),Innovation Program of Shanghai Municipal Education Commission(11YZ109)and the critical scientific and technological projects of Science and Technology Commission of Shanghai Municipality(11142200403).In this paper,the physicochemistry indexes were studied by principal component analysis to establish comprehensive indexes for evaluating sunflower oil quality during different oxidation processes.LF-NMR relaxation properties were used to built Prediction models,and the models were employed to forecast physicochemistry indicatorses and fatty acids content in sunflower oil.At last the correlation between LF-NMR relaxation properties and fatty acids composition in 36 kinds of fresh oils were analysedand oil classification by LF-NMR technology was verified.This study would be helpful to explain the relaxation mechanism of oils,and to prove the feasibility of monitoring oil quality by LF-NMR technology.The major results were as follows:(1)K232,K268,PAV,AV and TPC increased significantly(P?0.05)during the different oxidation processes,while the IV decreased and the POV under frying waved.After taking over the overall rate of change and correlations between the physicochemical indexes and UFA/SFA,POV or TPC,indicators(105?:POV,K232,PAV;180?: K232,PAV,TPC)were chosen to judge the sunflower oil quality accurately.Comprehensive indexes were achieved by PCA based on information fusion and the critical value is-0.189(105?)and 0.727(180?)respectively.The oil whose index was upon the critical value could be thought below standard.K232 and PAV could be used to distinguish oils oxidized in different temperatures well by means of PCA.(2)S21% increased binomially while T22 decreased and other physicochemical indicators fluctuated during the oxidation process at 105?.Both of S21% and T22 significantly(P?0.05)associated with peroxide value(POV),conjugated diene(CD),conjugated triene(CT)and anisidine value(AV)at 105?,while the similar significant correlations could be found between total polar compound(TPC)and indexes such as T21,S21%,S22% and T23.The models developed for POV,CD,CT,AV and TPC worked well(R2>0.85)at 105?,but it showed that only models for POV,CD and TPC had good forecast effect(R2>0.78).T21,T22,and S23% reduced during the frying process,while S21%,S22% increasedbinomially and T23 fluctuated.There were remarkable(P?0.05)correlations between LF-NMR ralaxation properties(T21,T22,S21%,S22% and S23%)and all physical and chemical indicators except POV at 180?.At the same time,only S22% correlated with POV significantly.All models used to predict physical and chemical indexes except POV was reliable(R2>0.89)and the accuracy was high(R2>0.85)after test.The results in this paper could be a reference for monitoring sunflower oil quality.(3)The fresh sunflower oil was mainly composed of linoleic acid(63.80%),oleic acid(29.48%),palmitic acid(5.73%),linolenic acid(0.54%),behenic acid(0.43%).The content of linoleic acid reduced significantly,while oleic acid and palmitic acid increased,and Linolenic acid and behenic acid fluctuated during different oxidation processes.The changes of fatty acids content increased with the rise of temperature,and differences of linoleic acid and palmitic acid content between 105? and 180?were significant(P?0.05).The unsaturation degree(MUFA/S and PUFA/S)of oils decreased when the anodization time extended.The principal component regression analysis showed that the model baded on LF-NMR relaxation properties could be used to predict fatty acids content well(R2>0.96)and the reliability of the model was verified(R2>0.94).(4)Both the fatty acids composition and LF-NMR relaxation characteristics of 36 fresh edible oils from 5 main categories(SFA,MUFA,MUFA/DUFA,DUFA,TUFA)according typical fatty acid was collected,and the correlation between themwere investigated.The results indicated that the LF-NMR relaxation signals of oils were composed of 3 peaks(P1,P2,P3)except palm oils.The relaxation characteristics such as T21,T22,T23 and S23%increased and S22% decreased while the unsaturation degree rised.The results of the principal component analysis(PCA)of LF-NMR relaxation characteristics indicated that the samples could be well categorized into the corresponding groups on the plot according totheir unsaturation of fatty acids.The influences of fatty acid unsaturation to LF-NMR relaxation characteristics were evaluated by stepwise regression analysis.It was found that P2 and P3 significantly correlatedwith SFA and PUFA respectively.The hydrogen protons which were not or a little impacted by the double bond were relevant to P2,while P3 was related to the hydrogen protons which were affected by the double bond.At last P1 was thought to have relevance to the minor constituent in oils such as tocopherol,sterols and oryzanol.To identify oil classification,discrimination analysis was conducted based on independent variable LF-NMR relaxation characteristics and dependent variable 1,2,3,4.The accuracy reached 95.83%while the test was 83.33%.The analysis results are credible.