The vegetable oil samples were prepared from the grape seeds of the varieties Vitis vinifera and Vitis labrusca, at approximately 100 Kg each. The seeds were collected during the harvest period (November 2010 to March 2011) from wineries located at two cities in the south of Brazil. The transportation was done in closed coolers to the Laboratory of Separation Processes of the Universidade Estadual de Maringá. Before oil extraction, the seeds were sieved out from waste, washed in deionized running water and then in hot water (∼100.0 °C) in order to remove husks, stalks and sugar as cited in Freitas (2007). After cooling down to room temperature, the seeds were packed in plastic bags and stored at -15.0 °C. For both extraction methods, the seeds were dried, respectively, at 40.0 °C and 80.0 °C.

In order to dry the seeds a convective drying with transversal flux was used. The seeds were dried at 40.0 °C and at 80.0 °C with an air flux speed of 0.8 m/s. The flux was controlled by means of a digital anemometer (TAVR-650). Before each oil extraction the seeds were ground in a grinder (model TE-345) for 10 seconds, homogenized and milled for another 10 seconds. When the desired drying was reached the mass of seed was measured using an analytical balance (GEHAKA-Bg 4000, 0.01 g of precision). The measurements of mass were made with an interval of one minute during 30 minutes. After this time the interval between two consecutive measurements of mass was two minutes. The humidity of the seeds was evaluated using Equation (1). X d b ( t ) = M ( t ) - M d M d

where X _db (t) is the humidity as a function of time in dry base, M(t) is the mass of seeds in the instant t, M _d is the mass of drying seeds.

In order to extract the oil two methods were used. One of them was the cold pressing and other one was the Soxhlet process. In the cold pressing a Bovenau press (capacity of 30 ton) was used. The extraction was made using 14 tons of pressure, during 12 h, on 100 g of grape seeds previously cleaned, dried and milled. The sizes of milled particles were found between 11.7 µm and 83.3 µm. The extraction cell was a stainless steel cell made in the mechanical laboratory of UEM (University of Brazil). In the Soxhlet process the solvent was dichloromethane and the extraction was made, for at least 4 h, at the melting point. In this method, the extraction was made with 10 g of seeds previously cleaned, dried and milled in 300 mL of solvent.

After extracting the oil from seeds dried at 40.0 °C and 80.0 °C the moisture content was determined using the Karl Fisher (Analyser, KF-1000) method. The relative humidity in oils was found between 0.22% and 0.52% considering both extraction methods, drying temperatures and types of grape. The maximum value was found for oil extracted by pressing V. vinifera seeds dried at 40.0 °C. The minimal value was found for oils extracted by pressing V. labrusca.

In order to determine the composition of fatty acids of the grape seed oil a gas chromatograph Varian, CO-3800 was used. To quantify the methyl ester, the standard methodology used by Visentainer and Franco (Visentainer and Franco, 2006) was followed. The standard methyl ester from tricosanoic acid (99%) was from Sigma-Aldrich. The fatty acid compositions of the oils extracted are shown in Tables 1 and 2.

A polarizing microscope (Leica DM 2500) with a camera (Leica DFC290) attached was used in order to analyze the birefringence of the oil samples. The POLM technique consists of observing a thin film sample between crossed polarizers. This is one standard technique applied in the investigations of anisotropic fluids such as liquid crystal (Mukai et al., 2007). When the light passes through the bire-fringent sample, it suffers modification in its polarization plane due to the optical anisotropy of the material, which in turn reveals the typical textures of each sample. If the sample is optically isotropic no light passes through the sample resulting in a dark texture. In this work the oils samples were placed in a capillar glass of 200 µm thickness. The optical analysis of the samples was made by rotating the platinum of the microscope in three different angles: 0°, 45° and 90° with respect to the polarization direction of the polarizer. The optical textures were obtained at room temperature and were observed by using the software Infanview.

Refractometry is an experimental technique widely used in the study of anisotropic fluids such as liquid crystals (Kimura et al., 2004) and composites formed by hydrogel doped with liquid crystal (Aouada et al., 2004). In this work, the refraction index of the oils was measured using an ABBE refractometer model Atago RX 5000α. The refractometer possesses a complete scale, ranging from 1.32700 to 1.58000 with a precision of 0.2% and a BRIX from 0–100.00% with a precision of 0.03%. The refractometer has a thermostat which allows for reading temperatures from 5.0 °C to 60.0 °C. In this work, the analysis temperature range used was from 16.0 °C to 50.0 °C. In order to avoid spurious impurities in the sample, care was taken with cleaning the refractometer. The cleaning was done with a solution of water and detergent (50%). This solution was applied onsite and removed carefully with a cotton swab to avoid risks of onsite measurements. After this, the excess solution was removed with a highly absorbant paper. This cleaning procedure was repeated three times. After the cleaning procedure, calibration was made by measuring the refractive index of distilled water at 20.0 °C. After calibration, 0.01 mL of oil sample was placed in the base of the refractometer and the refraction index of the oil was measured as a function of temperature.

A Solartron 1260A impedance-gain phase analyzer was used in order to measure the real R (f) and imaginary Im (f) parts of the electrical impedance Z (f) of the grape seed oils. The frequency range used was from 0.01 Hz to 1.0 MHz. The amplitude of the AC applied voltage was 1.0 V. The sample (V=15 mL) was placed between two circular plates of stainless steel with an area of 3.14 cm² and thickness of 0.1 mm. The thickness was controlled with a digital micrometer attached to the sample holder 12962 of the Solartron analyzer. The electrodes were cleaned with the following procedure: first, the electrodes were washed with detergent and deionized water. After that, the electrodes were polished with fine sandpaper. Finally, the electrodes were placed in an ultrasonic bath at 40 KHz for 10 min. The temperature sample was controlled with a thermostatically controlled water bath with a stability of 0.01 °C. The automated data acquisition was carried out by means of a computer equipped with ZPlot software from Solartron Analytical. In the frequency range from 0.01 Hz to 1 MHz, each value of impedance in their specific frequency is a simple mean of five consecutive measurements. In order to ensure reproducibility, the entire procedure was repeated three times with twenty minutes between each one. The reproducibility was obtained with a maximum deviation of about of 0.01%.

By using grape seeds of Vitis vinifera dried at 40.0 °C and 80.0 °C, the extracted oil content was 6% and 3% at each respective temperature when apply the pressing extraction method. When the Soxhlet process was applied, the content of extracted oil was 20% and 16% for the same drying temperatures. When the grape seeds of dried Vitis labrusca at 40.0 °C and 80.0 °C were used, the oil content was, respectively, 8% and 4% after applying the pressing method and 12% and 14% after applying the Soxhlet method. Each oil sample was analyzed using gas chromatography, and the presence of linoleic acid as the principal compound of the oil was confirmed. The percentage composition of fatty acids of the oils extracted for both extraction methods are shown in Tables 1 and 2 for drying temperatures of 40.0 °C and 80.0 °C, respectively.

Considering the deviation measurements in the chromatography technique there are no significant differences among fatty acids for both extraction methods. This shows that the temperature of seed drying and extraction method did not significantly influence the quality of the extract oil. All oils had high fatty acid contents (>81%). The chemical composition of the extracted oil contains essentially oleic, linoleic, palmitic and stearic fatty acids. This composition is in agreement with the mean composition of grape seed oil (Freitas, 2007). Table 3 shows the moisture contents of the grape seed oils extracted with pressing and Soxhlet methods at 40.0 °C and 80.0 °C drying temperatures. The values for moisture content in Table 3 confirm the expected results: increasing the drying temperature causes a decrease in the moisture content in the oil. For comparison, the moisture content in soybean oil, for example, is limited to up to 0.5% which is the maximum amount recommended so as not to harm the transesterification process.

In order to study the effective birefringence of the extracted oils by means of two different methods of extraction, the Polarized Optical Light Microscopy (POLM) technique was applied. Figures 1, 2, 3 and 4 show typical POM micrographs from the grape seeds oils of Vitis labrusca and Vitis vinifera when dried at 80.0 °C and extracted by pressing and Soxhlet methods. The optical textures, magnified by 100x, were obtained by placing the sample between crossed polarizers. The micro slide containing the sample was rotated in three different angles, 0°, 45° and 90°, with respect to polarization direction of the polarizer. It can be observed that the textures obtained for the oils extracted by pressing are typical of isotropic systems (Figures 1b and 1c) whose optical anisotropy (birefringence) is null. On the other hand, the optical textures of the grape seed oils extracted by the Soxhlet method show birefringent dots (identified by bright spots in Figures 2b and 2c). These birefringent regions are associated with the existence of anisometric particles in the oil. These particles are probably some few remains of the solvent used in the extraction by Soxhlet. In order to reduce the birefringent dots when using the Soxhlet method two ways are possible: increasing the drying temperature for both kinds of grape seeds or using Vitis vinifera grape seeds (see Figures 3 and 4) at any drying temperature used in this work. In this way, it is shown that the oils extracted from Vitis vinifera grape seeds are not influenced by the extraction method. Furthermore, a high drying temperature reduces the moisture in the seeds, thus contributing to the reduction in the amount of waste in the extracted oils.

The effective refractive indexes of the oils studied in this work were measured in an ABBE refractometer. Figures 5 and 6 show refractive index as a function of temperature, n (T), of the oils extracted from the seeds of Vitis labrusca (Figure 5) and Vitis vinifera (Figure 6) by means of Soxhlet (circles) and mechanical pressing (triangles) for seeds dried at 40.0 °C (black symbols) and 80.0 °C (gray symbols). The range of temperature used in these measurements was from 16.0 °C to 50.0 °C. As can be seen, n (T) decreases linearly with respect to temperature for both types of grape and for different drying temperatures of seeds. However, only with the pressing method is the linear behavior of n (T) reversible with temperature. By using the Soxhlet extraction method with Vitis labrusca grape seeds, a non-reversible behavior with respect to temperature and non-linearity of n (T) is observed (Figure 7). This non-linearity in n (T) can be associated with the presence of anisotropic particles in shape observed by the POLM technique. Anisotropic particles were not observed in the oils extracted by pressing. In general, one can see that the values of the refractive index for all the oils studied in this work are between 1.46200 and 1.47800. This value range of n (T) is higher than that obtained with grape seed oil extracted from the main Turkish wine grape cultivars (Baydar et al., 2007). In this work, the reproducibility of the measurements of n (T) was checked and deviations were less than 0.1%.

The real R (f) and imaginary Im (f) parts of the electrical impedance, Z (f), of the grape seed oils were performed using the EIS technique. Figure 8 shows the di-log plot of R (f) at 20.0 °C of the Vitis labrusca (v. Bordo) grape seed oil extracted by the mechanical pressing (open triangle) and Soxhlet methods (filled circle). Figure 9 shows the di-log plot of Im (f) at 20 °C of the Vitis labrusca grape seed oil extracted by the mechanical pressing (open triangle) and Soxhlet methods (filled circle). The seeds were dried at 40.0 °C. In both Figures 8 and 9 the continuous line is the best fit for the electrical circuit model of Figure 10 that will be discussed in section 3.4.1.

As expected in the frequency range from 10 mHz to 500 KHz, R (f) presents typically two plateaus connected with a linear decreasing between them. Some great differences have been observed between the Soxhlet and mechanical pressing extraction methods. The first one was the reduction of two orders of magnitude in the value of R (f) at a low frequency range when the Soxhlet method was used in comparison with mechanical pressing using the same grape. Another one is the shift in frequency of the maximum of Im (f) of one order of magnitude when the Soxhlet process is used. By fixing the Soxhlet method one can see that the kind of grape can produce a shift in the maximum Im (f) ranging from 200 Hz using Vitis labrusca (v. Bordo) to 400 Hz with Vitis vinifera grapes. The mechanical pressing does not produce significant shifts in the maximum of Im (f) when the type of grape is changed. As another difference between extraction methods is the width of first plateau of R (f) of two orders of magnitude lower when the Soxhlet is used. And, finally, as the last difference, a great reduction in the gap of the real part of impedance was observed when the Soxhlet method was used in the frequency region higher than 600 KHz.

Figures 11 and 12 show Nyquist diagrams for the oils (at 20.0 °C) extracted by Soxhlet (filled symbol) and mechanical pressing (open symbol) for both kinds of grapes (Vitis vinifera and Vitis labrusca) and for different drying temperatures. Continuum lines are the best fits obtained from the electrical circuit model (Figure 10). Figures 11 and 12 show clearly that, for both kinds of grapes, the real part of impedance decreases when the drying temperature is increased, and the oils extracted by mechanical pressing cannot be described by the simple RC circuit. This circuit model is in good agreement with the experimental data only for the oils extracted by Soxhlet. Table 4 show the best fit parameters for the circuit model applied in this paper. By using the dimensions of capacitor and capacitance from the best fit of model the electrical permittivity ɛ was evaluated. As can be seen in Table 4, only the values for the oils from V. labrusca seeds ɛ change drastically when the temperature for drying the seeds is changed. The ɛ of the oils from V. vinifera seeds have practically the same value when they are extracted by mechanical pressing.

Figures 13 and 14 show complex-plane plots of electrical impedance for temperatures from 16.0 °C to 50.0 °C of the grape seed oils extracted from Vitis labrusca (var. bordo) by mechanical pressing with seeds dried at 40.0 °C and 80.0 °C, respectively. Figures 15 and 16 show complex-plane plots for the oil extracted from Vitis vinifera (var. Cabernet) in the same conditions of preceding figures. With the exception of Figure 13, in which the oil was extracted by pressing from the seeds of Vitis labrusca dried at 40.0 °C, Nyquist diagrams for the mechanical pressing method has shown no semi-circle. An interesting result was noted in the complex-plane plots for both kinds of grapes for the oils extracted by mechanical pressing. All curves show a tendency toward the right side for seeds dried at 80.0 °C while for seeds dried at 40.0 °C this tendency goes to left side (see inset of Figure 15). This unexpected behavior can be associated with higher moisture content (see Table 3) in all the oils extracted by mechanical pressing. By using the Soxhlet extraction method the Nyquist diagrams show a semi-circle for both drying temperatures. For both extraction methods electrical resistance decreases with increasing temperature (see Figures 16 and 17).