3.3.1. Density
Table 3 shows the density values at different temperatures (10, 20, 30 and 40°C) for the sixteen VOOs. Density values were in a range from 0.904 to 0.915 g/cm3, depending on temperature. According to standard deviation values of different samples (sd < 0.001), and one-way analysis of variance performed, can be observed significant differences between densities of several VOOs varieties.
Table 3. Density values at different temperature for the sixteen VOOs (Standard deviations < 0.001 g/cm3).
Figure 2A illustrates the relationship between density and temperature for ‘Pi’, ‘Hj’, and ‘Ar’ VOOs, with a linear decrease for both variables according to Eq. (1) and similar to that found by other authors [7, 13].The values of the m and b were calculated for each VOO sample (Table 4). The parameters b and m were ranged from -3 x10-4 to -1 x 10-4 and from 0.913 to 0.917, respectively. According to these results, it is worth mentioning that temperature is a more important variable affecting oil density than the VOOs composition.
3.3.2 Dynamic viscosity
Figure 3A illustrates shear stress versus shear rate for the ‘Pi-I’ variety at three temperatures (10, 20, and 30°C). As can be observed, the shear stress is directly proportional to the shear rate and the slope corresponded to the dynamic viscosity. Results exhibited a Newtonian behavior, as described in previous works [8, 29]. The rheograms obtained for the other varieties were similar (data not shown), dynamic viscosity values for different temperatures are shown in Table 5.
Figure 3B shows oil dynamic viscosity as a function of temperature for the three selected varieties (‘Pi-I’, ‘Hj’, and ‘Ar’). As expected, a decrease in dynamic viscosity as a function of temperature was observed. All the samples exhibited the same dynamic viscosity pattern over temperature. In fact, although differences in chemical composition were observed, dynamic viscosity values seem to be similar for the rest of VOOs (Table 5). Various empirical expressions have been proposed relating dynamic viscosity to temperature [13, 14]. However, in this case, the Arrhenius model (Eq. (2)) describes the effect of temperature on VOO viscosity. The logarithm of the dynamic viscosity (ln p) versus the inverse temperature (1/T) (Eq. (3)) were plotted for ‘Pi-I’,
‘Hj’, and ‘Ar’ oils in Fig. 2B, showing a linear relationship which confirmed that Arrhenius model let to describe the experimental data:
The linear model parameters were estimated with the least square method with a good adjustment. Values of R2, ranging from 0.995 to 0.999 (Table 4), revealed that the Arrhenius model was efficient to describe the experimental data. The values of Ea and A were calculated for each VOO from slopes and intercepts of the Arrhenius plot, respectively (Table 4). Ea ranged from 30.65 to 33.63kJ/mol whereas A from 8.49 x 10-8 to 2.73 x 10-7 Pa ■ s. All these calculated values were quite similar to those found for four monovarietal VOOs by Bonnet et al. [8].
Figure 3. (A) Rheograms for ‘Pi-I’ VOO for three selected temperatures: 10, 20 and 30°C. (B) Evolution of dynamic viscosity (μ) as a function of temperature (T) for the three selected VOOs: ‘Pi-I’, ‘Hj’, and ‘Ar’.
3.4 Correlation between FA composition, thermal and physical properties
Pearson correlation coefficients (Table 6) were calculated among main FA (also grouped in SFA, UFA, MUFA, PUFA, and MUFA/PUFA) and thermal parameters of the cooling. As can be observed, in Table 6, enthalpy (AH) of crystallization was not influenced by chemical composition of VOO, since fatty acid composition was not correlated with this thermal property. On the contrary, all the other thermal properties obtained from the cooling thermograms were statistically correlated with FAs, whether free or grouped according to their unsaturation degree. In particular, C16:0, C18:1, C18:2, SFA, UFA, MUFA, PUFA, and MUFA/PUFA, exhibited highly
significant correlation with Pc, to, te, and R. However to showed a lower correlation with C18:2, PUFA and MUFA/PUFA. Similar results were previously reported by Chiavaro et al. [26] in Italian virgin olive oils, where enthalpy of crystallization was not influenced by the chemical composition, but all the other thermal properties obtained from the cooling thermograms were statistically correlated with the FAs, as well as TAG and lipid oxidation products.
Table 4. Coefficients of the (A) correlation between density and temperature and (B) Arrhenius model for the sixteen VOOs.
The dependence of the crystallization phase on oil composition was investigated correlating the temperature of the major peak from crystallization phase (Pc) with the FA composition. Two high correlations were observed between Pc and C18:1 (Fig. 4A) with r = 0.996 (p < 0.001) and C18:2 with r = -0.984 (p < 0.001). This significant fit demonstrates that some of the main FAs of VOOs (C18:1 and C18:2) are the main responsible of oil crystallization temperatures. A decrease in the crystallization temperature is related to a higher proportion of C18:2 and lower percent of C18:1. Similar results have been previously reported by other authors [16, 17, 33]. Jiménez et al. [16] found high correlation values (R2 = 0.95) between Pc and C18:1 and C18:2 content, as well as with SFA/UFA and MUFA/PUFA for six different VOOs. In the work were observed as high R values (range between to and te) were related to VOOs from varieties with high C18:2 content (‘Blanqueta’ variety), while that VOOs from varieties
with high C18:1 content, (‘Picual’ or ‘Hojiblanca’), showed a reduction in R value. Similar behaviors have been obtained in this study (Table 6), for the same varieties [16], as ‘Pi’, ‘Hj’, ‘Ar’, ‘Bl-II’, and ‘Bl-II’, and the other VOOs varieties studied with high C18:1 content (as ‘Kl’, ‘Me’, or ‘M.P’) show high Pc values and low R values. Those VOOs from ‘A.I’, ‘V.B’ or ‘T.L,’ with lower C18:1 content and simultaneously high C18:2 content, showed low Pc values and high R values.
Figure 4. Correlations between the oleic acid (C18:1) content of VOOs and the corresponding to (A) temperature of the major peak of crystallization phase (Pc) and (B) dynamic viscosities (u).
As for the cooling process, (Table 6), correlation coefficients (r) were also calculated among main FA (also grouped in SFA, UFA, MUFA, PUFA, and MUFA/PUFA) and thermal parameters of the melting process. However, in general were not found high correlations between thermal properties and FA composition. This can be explained considering that crystallization transitions were well-known to be more interpretable than those obtained upon heating, where melting-re-crystallization phenomenon, named polymorphism, could occur for the original oil crystals [19, 28].
Respect to the oxidation process, Jiménez et al. [16] established a high correlation between the time in which occurs tox and the oxidative stability (hours at 98°C) measured by Rancimat method. Monovarietal oils with greater oxidative stability had higher C18:1 content and oxidation time, while those with lower C18:1 content, with lower oil stability, had smaller oxidation times. For the oils analyzed high correlations (r) (Table 6) between tox and oleic acid (C18:1) were not found, and generally with none of the FAs. Other authors, Vecchio et al. [30], studied the thermoxidation of EVOO from different geographical proveniences (Italy, Spain and Tunisia) by DSC thermograms which were deconvoluted into constituent peaks, to obtain more information about the thermal composition of the EVOO relating thermal properties to chemical composition. In this way, were found correlations between the thermal properties of the peaks obtained by the deconvolution of the two events of the decomposition by DSC and the chemical composition, in particular with palmitic and oleic acids and related triacylglycerols.
The dependence of viscosity on the type of vegetable oil was investigated by several authors [8-10, 14] by correlating the oil viscosity with fatty acid composition (C18:1 and C18:2 fatty acids). In this study, although a non high correlation was found (Fig. 4B), viscosity tends to increase as the amount of C18:1 fatty acid. On the other hand, although a general decrease in dynamic viscosity was observed as C18:2 fatty acid lowered (data not plotted); relationships between both parameters (dynamic viscosity and C18:2) were not found. Thus, fatty acids with more double bonds do not have a rigid and fixed structure, being loosely packed and more fluid-like behavior [9]. These results demonstrated that the major components (C18:1 and C18:2 fatty acids) appear to make a great contribution to the VOO flow behaviors.
3.5 Multivariate analysis. Relation between FA composition, thermal and physical properties
The PCA multivariate was used to relate FA composition, thermal and physical properties for the different VOOs studied. The PCA multivariate technique has been performed using values of the parameters studied with a greater Pearson correlation: two main FAs (C18:1 and C18:2), two parameters from the DSC cooling analysis (Pc and R) and one of the physical properties studied (viscosity at 20°C), generating a score and loading plot (Fig. 5).
In fact, along the first-dimension PC1 (accounting for 90% of the total variance) were differentiated the monovarietal VOOs by thermal properties and chemical composition. VOOs from varieties with higher C18:1 content and high Pc values were plotted on the right side, while on the left-hand side were situated the VOO from
Figure 5. Score (A) and loading plot (B) of principal component analysis applied to the data set of main FAs (C18:1 and C18:2), parameters from DSC cooling analysis (Pc and R) and one of the physical properties studied (viscosity at 20°C), for the sixteen VOOs.
varieties with higher C18:2 content and high R valúes. On the other hand, VOO from varieties with high viscosity valúes were differentiated between the second dimension PC2 (accounting for 9% of the total variance), which were located at the top side of the plot. Comparisons between the two PCA plots put forward that ‘Bl-I’, ‘Bl-II’, ‘A.I’, ‘Fr’, and ‘Ar’ varieties were mainly discriminated by the C18:2 and R variables, whereas the ‘N.C’ ‘K.N’, ‘Hj’, ‘Pi-II’, and ‘Kl’ varieties were mainly separated by the C18:1 and Pc variables. The ‘T.L’, ‘V.Bv’, ‘Ma’, ‘Me’, ‘Pi-I’, and ‘M.P’ varieties were differentiated by the viscosity variable. Consequently, data obtained by the multivariate analysis confirmed those results previously obtained, showing that VOOs with high C18:1 content were related to high Pc values (Fig. 4A) and simultaneously to the viscosity values (Fig. 4B).In addition, VOO from varieties with high C18:2 content were related to high R values.
4. Conclusions
A first approach has been presented about the effect of FA composition and temperature on the physical (density and dynamic viscosity) and thermal properties (from crystallization, melting and oxidation processes) of VOO. As expected, oil density decreases linearly with increasing temperature for all VOOs and Arrhenius model described the effect of temperature on viscosity, showing a Newtonian behavior. Moreover, high correlation coefficients between thermal properties (temperature of the peak of crystallization phase, Pc) and FAs composition (C18:1 and C18:2) were found. Although non good correlations between oil viscosity and fatty acid composition could be found, the viscosity tends to increase as the amount of C18:1 fatty acid. These similar behaviors between main FAs (C18:1 and C18:2), parameters of the DSC cooling analysis (Pc and R), and physical properties studied (viscosity at 20°C) were also described by multivariate analysis (PCA). As a consequence, the information in this manuscript, both physical and thermal properties and its correlation with composition, can be considered to be important for many applications such as equipment design, storing and handling process to improve some aspects of the extraction process in olive oil mill. However, further research would be necessary to relate more compositional parameters (the location of double bonds, TAGs…), with physical and thermal properties of VOOs.
Acknowlegements: This work was supported by, a fellowship from Ministry of Science and Innovation (Spain) associated to the project FPI-INIA RTA2009-00002-00-0, and the project INIA-FEDER RTA 2010-00013-c02-01. The authors gratefully acknowledge their financial support. The authors have declared no conflicts of ¡nterest.