Olive oil is essential culinary oil of the Mediterranean diet. Olive oil is a genuine fruit juice with excellent nutritional, sensory and functional qualities obtained from the fruit of olive trees (Benito et al., 2013). Increasing evidence suggests that olive oil as a food has been associated with a lower incidence of chronic diseases, particularly coronary heart disease, a lower risk of cancer and other bioactive properties including anti-inflammatory, antioxidant, antiarrhythmic and vasodilatory effects (Covas Mi et al., 2006; López-Miranda et al., 2010). Many of the properties and reported health benefits of olive oil have been attributed to its minor components, such as sterols (β-sitosterol, Δ-5-avenasterol, etc), phenols (α-tocopherol, tyrosol, hydroxytyrosol, etc) and fatty acid composition (Hatzakis et al., 2010; Medeiros, 2001; Visioli et al., 2005).

Different categories of olive oil are legally defined by the European Community Regulation (EC, 2001), and are marked with different prices (Maggio et al., 2010). Thus, in order to increase profits there is the possibility of mixing seed oil or low-quality olive oil such as refined olive oil (ROO) and pomace olive oil (POO) with the highest quality product, extra virgin olive oil (EVOO). Besides infringing the benefits of consumers, the adulteration of extra virgin olive oil may cause serious safety problems. Olive oil adulteration resulted in over 20,000 people becoming sick and over 400 people died in 1981. This incident is called “Spanish toxic oil syndrome” (Jimeno, 1982; Tay et al., 2002). With economic growth, people in China are consuming more and more EVOO. At the same time, consumers and governments in China have paid a lot of attention to its quality issues. For the past 10 years the percentages of diacylglycerols (DGs) in olive oil have been investigated for their usefulness as a marker of possible adulteration (Pérez-Camino et al., 2001). Several studies suggest that the content of the percentages of total DGs and the ratio of 1,2-DGs to the total amount of DGs are useful indices to assess the freshness and quality of olive oil (Fronimaki et al., 2002; Pérez-Camino et al., 2001; Sacchi et al., 1991).

Over the past few years, NMR spectroscopy has been widely used for the analysis of virgin olive oil to evaluate quality and authentication. In particular, ¹H and ¹³C NMR spectroscopies have been successfully applied to the study of minor components in virgin olive oil, and have revealed how these components are related to oil quality (Sacchi et al., 1996; Sacchi et al., 1997). ³¹P NMR spectroscopy is a convenient method which was introduced to supplement ¹H and ¹³C NMR spectrocopies, especially in cases where overlapped peaks in the ¹H NMR spectra or long relaxation times of the insensitive ¹³C nuclei make the analysis a difficult task (Spyros and Dais, 2000). The ³¹P NMR spectrum of virgin olive oil can illustrate the region of the phosphitylated total sterols, monoglycerides, DGs and free fatty acids (Vigli et al., 2003) by applying a method which is very fast, easy and clear, and can determine the quality of olive oil effectively.

As early as a few decades ago, a number of ¹⁹F NMR reagents have been studied to readily react with various active hydrogen functional groups. Among these reagents, p-fluorobenzoyl chloride has ever been used to characterize alcohols, phenols, carboxylic acids, amines, or thiols qualitatively by fluorine-19 nuclear magnetic resonance spectrometry (Spratt et al., 1984).

In this paper, we developed a completely new application of ₁₉F NMR spectra for determining the quality of olive oil because the deriving reagent for ₃₁P NMR, phosphorus trichloride, has been banned in China since 2008 due to its high toxicity. The primary task of this work was to test this safer technique with certain model compounds bearing labile protons and to determine these compounds quantitatively, then apply it to the determination and analysis of olive oil samples.

This method is based on the derivatization of the labile hydrogens of groups, such as OH and COOH, of different types of compounds with 4-fluorobenzoyl chloride according to the reaction shown in Fig. 1 and the integration of the appropriate peaks in the ¹⁹F NMR spectrum.

The natural abundance of ¹⁹F is as the same as ³¹P (100%), much higher than ¹³C (1.11%), and a little higher than ¹H (99.985%), so ¹⁹F NMR is sensitive and can give strong signals. The applicability of this method to the qualitative determination of these model compounds bearing active functional groups is demonstrated in Fig. 2, which illustrates the ₁₉F NMR spectrum of acids, a mixture of some model compounds, glycerol, 1,3-DGs and 1,2-DGs. Well-separated peaks for these model compounds are observed. The peak at δ −164.90 ppm and a peak at δ −107.47 ppm belong to hexafluorobenzene and the internal standard 4-tert-butylphenol, respectively. In the ¹⁹F NMR spectrum (Fig. 2a), the intense peak K at δ −105.05 ppm is attributed to the deriving reagent. The broader, but smaller peak J at δ −103.57 ppm is attributed to the intermediate product in the reaction between pyridine and the deriving reagent. It does clearly show that two groups of peaks which belong to the derivative of glycerol separate very well (Fig. 2c). The α- and β-hydroxyl groups of glycerol are located at −107.89 ppm and −107.71 ppm, respectively. Moreover, the integral area of the α-hydroxyl groups is exactly twice the β-hydroxyl group.

Table 1 summarizes the ¹⁹F NMR chemical shifts in the derivatives of the model compounds used in this study. Table 1 reveals a number of interesting features. First, the aliphatic hydroxyl groups are clearly distinguishable from the aromatic and other enolic hydroxyl groups in the spectra of the derivatives of the model compounds. The chemical shifts of the two derived hydroxyl groups of tyrosol can be the best illustration. The derived phenolic hydroxyl of tyrosol appears at δ −107.32 ppm and that of the aliphatic one at δ −108.58 ppm. The integral areas of derived aliphatic hydroxyl and derived phenolic hydroxyl of tyrosol are equal, which means that the aliphatic hydroxyl group and phenolic hydroxyl group can both equally react with 4-fluorobenzoyl chloride. Second, the ¹⁹F chemical shifts of these model compounds containing –COOH and -OH are located in different fields. Obviously, the chemical shift of the derived carboxylic acid used in this study appears in a lower field than that of the other model compounds. Third, the ¹⁹F chemical shift is sensitive to the position of the hydroxyl in 1,3-DG and 1,2-DG. Based on the chemical shifts of acids and alcohols in Table 1, it was found that several common fatty acids have the same ¹⁹F NMR chemical shift, and so do several common fatty alcohols. It can be inferred that the acids and alcohols containing secondary carbons have the same chemical shifts. In our study, we found ethanol appeared to be at δ −108.93 ppm, the same as aliphatic alcohols, while isopropanol appeared to be at δ −109.13 ppm the same as cyclohexanol, which can well demonstrate the above inference. In conclusion, the relatively small size of the fluorine nucleus (between those of hydrogen and oxygen), a nuclear spin of 1/2, 100% natural abundance, high NMR sensitivity (83% of that of a proton) and pronounced long-range effects (Gakh et al., 2000) facilitate a clear distinction of these model compounds.

It was necessary to test the ability of ¹⁹F NMR for quantitative analysis. The internal standard 4-tert-butylphenol was used for the quantification of various compounds. A comparison of the NMR data with a known amount of the typical compounds at different ratios for ¹⁹F NMR is shown in Table 2. As can be seen, the values of relative tolerance range from 0.5 to 6.5%. Particularly, the relative tolerance for the compound of hexadecyl alcohol is the smallest, indicating that hexadecyl alcohol reacts more easily and quantitatively with the deriving reagent (4-fluorobenzoyl chloride) than other typical compounds.

Checking and comparing the ¹⁹F NMR spectra of glycerol and tyrosol, primary alcohol, secondary alcohol and phenols without hindered substituents on 2,6-positions can quantitatively react with the deriving reagent. In our research, fatty acids were not shown to react with the deriving reagent quantitatively like alcohol.

37 Commercial olive oil samples from different countries and 5 other plant oil (PO) samples purchased from the local supermarket were derived with 4-fluorobenzoyl chloride, and their ¹⁹F NMR spectra were recorded. A typical spectrum is shown in Fig. 2f. In the ¹⁹F NMR spectrum, the two more intense peaks at δ −107.86 ppm and −107.95 ppm are attributed to 1,3- and 1,2-DGs, respectively. The peaks with Arabic numbers in the spectra have been assigned as follow: 1, free fatty acids; 2, unknown. 4-Tert-butylphenol was apparently a suitable internal standard because of its peak being well-separated from other components of olive oil and rapid reaction under conditions with the deriving reagent. The peaks, Xes appear at the same chemical shifts in the blank sample and other oil samples (Fig. 2f), which are impurities of the deriving reagent.

The integrals of these peaks are used to determine 1,2-DGs, 1,3-DGs, total DG content, and ratio D (1,2-DGs/total DGs) quantitatively. The results of such an analysis for the virgin olive oil from various countries and some plant oils are summarized in Table 3. The total amount of DGs in all VOO samples ranges from ~1.5 to ~9%. EVOO samples and blended olive oil (BOO) samples show an average mean value of total DGs of 2.51% and 5.67%, respectively. The value of total DGs of the EVOO samples is similar to Fragaki et al. (2005). Obviously, the total DG content in the EVOO samples is lower than that in the BOO samples which agrees with the statement that total DG concentration increases during the storage of olive oil (Fronimaki et al., 2002). The majority of the samples appear to contain a higher content of 1,3-DGs than 1,2-DGs which is contrary to Fragaki et al. (2005). Only sample 2 from Spain appears to have a lower amount of 1,3-DGs. The contents of total DGs and the ratios (D) of 1,2-DGs to the total amount of DGs are two important parameters to differentiate the olive oil samples. Previous research on the ratio (D) of virgin olive oils freshly extracted from olives of normal ripeness should be close to 1 according to Vigli et al. (2003). The ratios D of all samples we studied are lower than 1 which can be explained by the fact that all the samples used in our work are imported and have been stored for a long time. The ratios D in all EVOO samples range from ~0.25 to ~0.5. The EVOO samples from Spain show the highest ratios, followed by Italy and Greece. The ratios D of BOO samples are commonly lower than 0.3. For virgin olive oil, the amount of DGs and the ratio D depend on several factors, such as the olive variety, the ripeness of the olive fruit, the environment, and the storage life of the product (Fronimaki et al., 2002). As a result, the analysis of the amount of DGs and the ratio D can be used to monitor different grades of olive oil. In addition, the 1,3-DGs content of all the five plant oils is higher than 1,2-DGs and the ratio D is ~0.3 which are in agreement with Vigli et al. (2003).

Figure 3 shows the ratio D against the total DGs for the samples of the extra virgin olive oils and the blended olive oils from different countries. Eight extra virgin olive oil samples (samples 1,2,3,4,5,8,10,11) are clearly gathered at the upper left of the plot (high D values and low total DGs), indicating that they are high-quality olive oil. According to EEC Standard the free fatty acid (FFA) content cannot exceed 0.8%. In theory, one mol DG (based on diolein molecular weight of 621) will produce one mol FFA (based on oleic acid molecular weight of 282) if one mol triglyceride is hydrolyzed, so 0.8% FFA will produce 1.8% total DGs. If other factors are considered, 2.5% DGs is a reasonable parameter. The other 16 samples labeled as extra virgin olive oil with lower ratio D values (<0.35) or higher total DGs content (>2.50%) tend to fall in the lower left corner, a fact that can be explained by long shelf life or adulteration with refined olive oil or even other refined vegetable oils. The blended olive oil samples (low D value and high total DGs) tend to cluster at the lower right of the plot (Fig. 3). Interestingly, the position of sample 16 falls in the same region of the plot with the BOO samples. The lower limit of the ratio D value and the upper limit of total DGs in extra virgin olive oil are 0.35 and 2.50%, respectively. As a result, it can be deduced that the samples at the upper left corner of the plot can be classified as extra virgin olive oils while the samples at the lower right corner of the plot can be classified as blended olive oils or refined olive oils.

Comparing the other 5 plant oils with olive oils, two of them have a lower DG content than 2.5%, but all of them have lower D values than 0.35. Nowadays, most EVOOs are adulterated with refined olive oils. If other plant oils, with the exception of tea seed oil, are added to olive oils, their oleic acid content will reduce and can be easily detected by GC or ¹H NMR.

The study demonstrates that the ¹⁹F NMR technique has great potential to detect and analyze the quality of virgin olive oil. Its high sensitivity, excellent resolution and short detection time (~5 min) allow it to determine more minor components in olive oil. Moreover, no decomposition of products is observed for several hours at room temperature after derivatization. This method is an easier, simpler, faster, safer and more reliable technique for comparing the classic analytic methods such as HPLC, GC and chemical analysis for the determination of the quality of olive oil. This method can also be extended to monitoring the quality of ordinary edible oils. Further study is in progress in our laboratory.