Nine different fillings were produced in a local confectionery (Sugar Bloom/Mestre Cacau, Beja, Portugal), using white chocolate (SicaoCDW-U2630-557), cream (30% fat) and three different references of olive oil at three different concentrations (20%, 25% and 30% w/w), as described in Table 1. The preparation of fillings started by heating the cream up to 100 °C, using a heating plate provided with a magnetic stirrer. After that, the heating plate was turned off and white chocolate and olive oil were added to the mixture using a double-blade hand blender to homogenize, according to the percentages presented in Table 2. A control sample was also prepared, with no olive oil. Finally, the fillings were kept at 25 °C for twenty-four hours to totally crystallize the cocoa butter included in the white chocolate.

The olive oil reference “ArbD” started with a bleaching process of virgin olive oil “Arb”. Virgin oil “Arb” was stirred with 2% w/w bleaching earth SepiGel Natural 200RF (Sepiolsa) for 120 min at 60 °C. At the end of the bleaching process, the mixture was filtered using a Macherey-Nagel MN321 paper filter (7–12 µm) and stored at 20 °C in an Erlenmeyer flask covered with aluminum foil.

The pH was evaluated at 20 ± 1 °C using a Metrohm 691 pH Meter (Switzerland). Moisture content was evaluated using method 931.04 (AOAC, 1990). Determinations were made five times.

The viscosity measurement of olive oil was carried out with a VT 550 rotational viscometer (Thermo Haake, Germany), using the concentric cylinder MV-DIN at 20 °C. The tests were conducted with a controlled shear rate (from 200 to 1000s⁻¹). The obtained flow curves were evaluated according to the Newton model. Determinations were done in quintuplicate. The flow curves of the chocolate fillings were carried out with a VT 550 rotational viscometer (Thermo Haake, Germany) using cone-and-plate geometry at 20 °C. The measuring device was PK1 1° (10 mm Ø). Shear stress ranged from 10 to 1000 s⁻¹. Water was pumped into the jacketed vessel of the viscometer (Grant LTD6G) to control the temperature of the samples at 20 ± 1 °C during measuring. The obtained flow curves were evaluated according to the rheological model of Ostwald, referred to also as the Power law model and used in previous studies on chocolate (Dias et al., 2015). Determinations were made in quintuplicate. This model is represented in Eq. (2):

σ = K γ ⁿ    (2)

Where σ is the shear stress (Pa), K is consistency index (Pa.sⁿ), γ is the shear rate (s⁻¹) and n is the dimensionless flow behavior index.

Small amplitude oscillatory measurements of chocolate fillings were performed using a controlled shear-strain rheometer (Malvern Kinexus lab+, England) connected to a refrigeration circuit with controlled temperature. The visco-elastic behavior of the fillings was evaluated at 20 °C using a 4°/40 mm cone and plate geometry and gap distance of 1 mm. First, excess sample was trimmed with a thin plastic blade and the sample was allowed to rest for 5 min for temperature equilibrium. Later, the linear visco-elastic region (LVR) was evaluated by performing a strain sweep (0.001%–1000%) at a steady frequency of 1Hz. Finally, the dynamic frequency sweep was conducted applying a steady strain of 0.01%, within the LVR, from 0.001Hz to 100Hz. The rheological parameters for this study were the storage (G’, in Pa) loss (G’’, in Pa) moduli.

Average, standard deviation and 0.95 confidence interval values were determined. Experimental data was subjected to one-way ANOVA (pair-wise comparison of means with the Scheffé test) in order to find differences in samples with different concentrations of olive oil. The data was analyzed using Statistica 6.0 (Statsoft, Tulsa, USA). The principal component analysis (PCA) was carried in order to evaluate the importance of the use of olive oil in the production of the fillings (Alvarenga et al., 2011).

The spectral power distribution of olive oils GC and Arb were similar and both presented lower transmittance values than ArbD in every wavelength. Both (GC and Arb) presented a maximum transmittance peak at 580–600 nm followed by a decrease to 610 nm (Figure 3). In both cases the minimum valued was recorded at 670 nm, identified as the spectral range with the highest absorbance of pheophytin“a”, the major fraction of chlorophyll (Fuentes de Mendoza, 2013), and usually used to evaluate the chlorophyll content of olive oil (Morelló et al., 2003). On the other hand, wavelength 470 nm was identified as the wavelength with the highest absorbance by lutein, the major fraction of carotenoids (Morelló et al., 2003) presenting antioxidant activity and recommended to prevent some types of cancer (Limón et al., 2015). These results indicate that olive oil GC presented the highest contents in chlorophyll and carotenoids, followed by Arb and finally ArbD.

Table 3 presents the estimated tristimulus (XYZ) and CIE 1976 L*a*b* values, based on the spectral power distribution. The tristimulus values presented significant differences (P<0.05) among the three olive oils, according to the visual perception. The highest values were recorded for olive oil ArbD (lighter appearance) and the lowest were recorded for olive oil GC (darker appearance). Also, parameter L* from CIE 1976 color space was the highest for ArbD (L*=94.8), followed by Arb (L*=78.7) and, finally, GC (L*=56.5). Similar results for the Arbequina variety were observed by Moyano et al. (2008). The observed L*a*b* values were consistent with other studies on color analysis of olive oil, especially L* and b* parameters (Moyano et al., 2008; Pérez et al., 2003). The values of Arb were similar to the results of olive oils from the Extremadura region (Spain) produced with Morisca and Carrasqueña varieties (Fuentes de Mendoza, 2013). Parameter a* presented the highest values for GC olive oil (a*=3.6), followed by Arb (a*= – 6.4) and ArbD (a*= – 10.7). Parameter b* presented positive values in all samples and the results of GC and ArbD were similar to olive oils using the Morisca and Carrasqueña varieties (Fuentes de Mendoza, 2013).

The viscosity of olive oil ranged from 61.03 mPas (ArbD) to 92.26 mPas (Arb), similar to other results on virgin olive oils at 20 °C (Gila et al., 2015). The lowest viscosity value was observed in the bleached olive oil ArbD (61.03 mPas).

The moisture values of chocolate fillings ranged from 8.3% (A20ArbD) to 13.1% (A25Arb), as a consequence of the added cream. As shown in Table 4, the results of moisture content of formulations using olive oil presented higher values than the control sample (P<0.05), except for sample A20ArbD. However, no differences were expected due to the same cream content in all formulations and the same moisture content both in the white chocolate and the olive oil (<1%w/w) used in the preparation of the samples. The results of moisture content were lower compared with other studies, due to the lower cream content (20% cream w/w), instead of the common value 30–40% cream w/w (Wybauw, 2010). The pH values were lower in the control sample (pH=5.79). The formulations using olive oil presented pH values from 5.91 (A25GC) to 5.97 (A25ArbD, A30ArbD), but no correlation was observed between added olive oil and pH (P≥0.05). Figure 4 presents the results of all chocolate fillings on the a* and b* components of the CIE 1976 color space. Based on the olive oil, the identification of three main clusters is noticeable: GC (lower values of a* and higher values of b*), Arb (central values of a* and b*) and Control/ArbD (higher values of a* and lower values of b*). The results of the colorimeter for the chocolate fillings presented higher L* value for the control sample (L*=82.0), with no addition of olive oil, and all results presented a negative correlation between added olive oil and L* (Table 4). This influence was most noted using GC olive oil, followed by Arb olive oil and less noted using ArbD olive oil, as a consequence of the L* of the olive oil (Table 3). Both parameters a* and b* were influenced by the type and concentration of olive oil. And, as expected, the highest shift from the control sample was observed when using GC olive oil at 30% w/w.

As mentioned before, luminescence (Y) was calculated based on the RGB data obtained from the digital image analysis and values ranged from 0 (black) to 255 (white). The highest value for luminescence was observed in the control sample (Y=225), followed by Arb and ArbD olive oil, with no significant differences (P≥0.05). On the other hand, the lowest value was obtained in the chocolate fillings produced with GC olive oil, presenting a negative correlation with olive oil (P<0.05). The results of the digital image analysis were consistent with the colorimeter results and both Y and L* presented higher values in the control sample and lower values in the sample coded A30GC.

The results of the image and color analyses conclude that ArbD is the olive oil with the best ability to maintain the original appearance of conventional chocolate fillings and, on the other hand, GC olive oil considerably changes the fillings to a more yellow-green color. Nevertheless, sample A20GC, due to the lower olive oil content did not affect color by as much as 25% and 30% w/w.

The Power Law model was used to evaluate consistency (K) and flow index (n). The results showed that all formulations presented shear-thinning behavior. Nevertheless, a significant gap was observed between the control sample (n=0.01) and formulations using olive oil (0.13<n<0.25). As expected, the results of flow index presented a negative correlation with consistency, as reported in previous studies on chocolate beverages (Dogan et al., 2011; Yanes et al., 2002), low-fat chocolate fillings (Dias et al., 2015) and low-fat dairy desserts (González-Tomás et al., 2008). The highest value of consistency (K) was observed in the control sample (K=1371 Pasⁿ) and a significant decrease due to olive oil was observed in all formulations (Table 4), probably due to higher content in unsaturated fatty acids in virgin olive oil (Gila et al., 2015) compared with cocoa butter (Torres-Moreno et al., 2015). These results are consistent with previous studies reporting the relation between lower solid fat content and lower chocolate viscosity (Rothkopf and Danzl, 2015; Talbot, 2009), where the main identified causes were the increase in liquid content, TAG with a different nature and a different chain length from cocoa butter (Rothkopf and Danzl, 2015).

Although differences were observed in the viscosity of the tested olive oils (Table 3), no influence on the consistency of chocolate fillings was observed.

The results obtained from the rheometer (Table 4) showed that the control sample presented a higher value for the storage module G’_1Hz (4.42x10⁶Pa), than the fillings using olive oil (0.28x10⁶Pa to 2.51x10⁶Pa). These results are coherent with consistency K (Table 4) and reflect the importance of cocoa butter in the structure of chocolate products (Dias et al., 2015; Afoakwa, 2010). The use of olive oil caused a decrease in storage module G’_1Hz, compared with the control sample, even using an olive oil concentration of 20% w/w, reflecting a less compact structure and a more liquid-like behavior (Di Mattia et al., 2015). The decrease in the storage module is a consequence of the olive oil composition, especially the larger content in unsaturated fatty acids and the different composition of triacylglycerols, compared with cocoa butter (Boskou, 2011). On the other hand, chocolate is a suspension of solid particles (like sugars, milk proteins, cocoa solids), around 30-50 µm in diameter, in a continuous fat phase, mainly cocoa butter (Beckett, 2008). The addition of olive oil will contribute to the increase in continuous fat phase, thus decreasing the proportion of solid particles and, as a consequence, lower values of viscosity and G’.

The results of the ANOVA post-hoc comparison (Table 4) did not present an influence of the type of olive oil on the rheological behavior (P≥0.05). Nevertheless, the results of the principal component analysis (Figure 6) presented differences in the fillings produced with ArbD (lower G’_1Hz) and GC (higher G’_1Hz), which is consistent with the results on the viscosity of the olive oils used (Table 3).

The frequency dependence of the dynamic moduli was evaluated and data from the storage module (G’) and loss module (G’’) are shown in mechanical spectra (Figure 5). The results of G’ and G’’ for the control sample presented limited dependency with frequency. In addition, the storage module (G’) presented higher values than the loss module (G’’), especially from 0.01Hz to 100Hz. These results are coherent with a gel structure. The use of olive oil caused a decrease in both storage and loss moduli and increased the dependency on frequency, especially between 0.001 and 0.1Hz. The presence of a crossover between dynamic moduli (i.e. G’=G’’) was observed at 30% w/w olive oil and, as a consequence, can be assumed as the critical value defining the transition between a suspension and a gelled system (Lupi et al., 2013).

A principal component analysis (PCA) was carried out with the color parameters L* a* b*, two chemical parameters (moisture, pH) and three rheological parameters (G’_1Hz, K, n). Figure 6 shows the PC1 vs. PC2 projection of samples (average values) and, in the same plot, orientation of the parameters in the plot. The similarity map defined by the first two principal components took into account 81.2% of the total variance. The first component (PC1) by itself condensed 58.7% and the second component (PC2) represented 22.5% of the total variance. The PC1 presented negative correlations with the rheological parameters (G’_1Hz, K) and L* parameter and a positive correlation with the n flow index and pH. The PC2 was negatively correlated to the b* parameter and positively correlated with a*. Consequently, the results of the multivariate analysis confirmed the results previously obtained, showing the softening effect of olive oil over texture and the effect of GC olive oil on the darker color of chocolate fillings (Figure 6).