In the present work, ethanol (99.75%, JT Baker) and n-hexano (98.5%, Anidrol) were used as solvents and the oil was obtained from the seeds of passion fruit (purchased in the local market of Umuarama, PR). Methyl heptadecanoate (> 99%, Sigma-Aldrich), 5α-cholestane (> 99%, Sigma-Aldrich), a FAME (fatty acid methyl esters) mixture (Supelco), heptane and the derivatization reagent Boron trifluoride-methanol solution (14%, Sigma Aldrich) were used for the characterization of the oil, and N,O bis (trimethylsilyl) trifluoroacetamide (1% TMCS, Fluka). Novozym® 435 lipase (Candida Antarctica immobilized), heptane (Anidrol) and methyl heptadecanoate (Sigma-Aldrich, 99.9% purity) were used to determine the ethyl ester contents in the samples, and N-Methyl-N- (trimethylsilyl) trifluoroacetamide (MSTFA) was used to determine the mono-, di- and triglyceride contents.

The fruits were cut in half in the horizontal direction, and their seeds were removed manually. The seeds were then washed under running water, the surface water was removed, and the seeds were put in the oven (Marconi, MA 035) at 70 ºC to obtain samples with a moisture content of 2.34 ± 0.69%. After drying, the aryl was removed completely and the material was triturated and passed through granulometric grading to separate the 0.638-mm fraction for the experiments.

UAE was conducted using ethanol as a solvent. In each experimental run, the flask containing the passion fruit seeds and ethanol was placed in the center of an ultrasound bath (Ultronique, Q 5.9/40A), previously heated to the experimental temperature, and ultrasonic irradiation was immediately initiated (power of 165 W and frequency of 25 kHz). To avoid the loss in solvent, the flask was connected to the condenser which was held at 10 °C.

In order to determine the effect of the process variables and the experimental conditions that maximize the removal of the oil from the seeds, the following levels for each variable were evaluated: (a) times (X₁) of 10, 30, and 50 min; (b) temperatures (X₂) of 30, 45, and 60 ºC; and solvent-to- seed ratios (X₃) of 2, 3, and 4 (mL·g^-1), which were combined by means of a Box-Behnken design. The experimental range for process variables was defined based on previous studies: time (Pereira et al., 2017), temperature (Rodrigues et al., 2017; Rosa et al., 2019; Xu et al., 2016) and solvent-to-seed ratios (Oliveira et al. 2014).

Analysis of variance was carried out to evaluate the effect of the variables on the oil yield with a 95% confidence interval using Statistica 8.0 software (StatSoft^TM, Inc.). In addition, the experimental data were fitted to a second-order polynomial model, which was used to determine the condition of maximum oil extraction. In this condition, the effect of ultrasound on the removal of oil from passion fruit seeds was evaluated, with extractions performed without ultrasound in an orbital shaker (Marconi, MA 839/A) at 40 rpm as described above.

For comparative purposes, the traditional extraction was performed in a Soxhlet extractor using the n-hexane a at a sample ratio of 30 (mL·g^-1) for 480 min.

For each extraction method used, after the duration of the experimental run, the seeds were separated and the solvent was removed until constant weight was achieved, thereby obtaining the oil. The oil yield was determined as the ratio between the mass of oil extracted and the mass of seeds used.

The methodology reported by Gonzalez et al., (2013) was used in the preparation of the samples to determine the fatty acid composition. In this methodology, the samples were derivatized with BF3-methanol and subsequently methyl heptadecanoate diluted in heptane was added to quantify the compounds. The analysis was performed in a gas chromatograph (GC-MS QP2010 SE, Shimadzu) equipped with Rtx - Wax (Shimadzu, 30 m x 0.32 mm id x 0.25 µm) capillary column and flame ionization detector, using the chromatographic conditions reported by Stevanato and Silva (2019). The components present in the samples were identified through comparison with authentic standards of fatty acid methyl esters.

In the determination of phytosterol and tocopherol contents, 5α-cholestane as internal standard and thereafter 40 µL of N,O-bis (trimethylsilyl) trifluoroacetamide were added to 40 mg of sample. The samples were maintained for 60 min at 60 °C and after dilution in heptane to obtain a volume of 1000 µL of solution, they were analyzed on an GC-MS QP2010 SE (Shimadzu) with SH-RTx-5MS column (Shimadzu, 5% phenyl-methylsiloxane, 30 m × 0.25 mm id x 0.25 μm). The column temperature was initially programmed at 50 °C, increasing to 230 °C at a rate of 10 °C·min^-1 and then to 280 °C at a rate of 15 °C·min^-1, which was held for 22 min. Helium was used as the carrier gas at a 1.0 mL·min^-1 flow rate with split ratio of 10 and the amount of injected sample was 1 μL. Mass spectra were recorded at 70 eV with mass range from m/z 50 to 550 amu. Compounds were identified through the comparison of spectrum data to those presented in the NIST14.lb and NIST14.lbs spectral libraries.

The UAE product (oil + ethanol) was added again in an Erlenmeyer flask and received the enzymatic catalyst in the proportion of 40% (relative to oil mass) as defined by Stevanato and Silva (2019). The enzyme was activated for 1 h at 40 °C prior to use. The flask was incubated in orbital agitation (Marconi MA 830/A) at 150 rpm by applying reaction times of 12 and 24 h. The experiments were carried out under the temperature of maximum oil yield determined for the UAE (60 ºC). At the end of each reaction, the enzymes and seeds were separated by filtration and the excess ethanol in the samples was evaporated until reaching constant weight.

The analysis of the chemical composition of the reaction product was carried out in a gas chromatograph (Shimadzu, GC-2010 Plus) equipped with a flame ionization detector and capillary columns SH-Rtx-Wax™ (Shimadzu, 30 m × 0.32 mm × 0.25 μm) and ZB-5HT inferno™ (Zebron, 10 m × 0.32 mm × 0.10 μm) to determine the content of methyl esters and acylglycerols, respectively. The chromatographic conditions for both analyses were previously established (Stevanato and Silva, 2019).

To quantify the content ethyl esters, the injected sample was prepared with 100 µL of a solution of oil in heptane (C = 30 mg·mL^-1), 80 µL of internal standard (methyl heptadecanoate, C = 12.5 mg·mL^-1) and 820 µL of heptane. For simultaneous determination of the contents of glycerol, mono-, di- and triglycerides, MSTFA was used as a silylating agent. Glycerol, monolein, diolein and triolein were used as external standard to construct the calibration curve (R² > 0.99).

The free fatty acid (FFA) content in the reaction product was analyzed in a gas chromatograph (Shimadzu, GC-2010 Plus) and the samples were prepared and analyzed according to the methodology for determining phytosterols and tocopherols as described in section 2.4, using methyl heptadecanoate as an internal standard.

The data were subjected to ANOVA using Excel® 2010 software and the Tukey test (p=0.05) to evaluate differences between the results. The experiments and analyses were performed in duplicate, so four observations were used to calculate each mean value.

Extraction performed without the use of ultrasound, provided an oil yield of 15.82 ± 0.8% under the conditions of maximum oil yield obtained by Equation 1, which represents 75% of the yield obtained with ultrasound. The advantages and characteristics of ultrasound have already been discussed and justify the result obtained.

Table 3 shows the results of the comparison between traditional extraction and UAE in terms of oil yield, fatty acid composition, and phytosterol and tocopherol contents.

In the traditional extraction (in Soxhlet), the passion fruit seeds used in this work presented oil yields of 29.02 ± 1.61%. This result is close to that reported in the works of Oliveira et al., (2013), Pereira et al., (2017) and Santana et al., (2017), who presented oil yields of 26.4, 28.33, and 28.87%, respectively.

UAE provided 73.7% of the yield obtained from traditional extraction; however, with this method the oil is obtained in a shorter time and mainly using a smaller volume of solvent, since a solvent-to-seed ratio of 4 (mL·g^-1) is required, while in traditional extraction a solvent-to-seed ratio of 30 (mL·g^-1) is used. In UAE, the cavitation of the medium favors mass transfer and thus a smaller volume of solvent is necessary; whereas in the classic method the extraction process is governed by mass diffusion, requiring a high consumption of solvent (Menezes et al., 2018).

The results presented in Table 3 show that there is no difference between the fatty acid compositions of the oils obtained with the techniques used. Passion fruit seed oil contains a high concentration of PUFAs (59.28 to 61.19), with linoleic acid forming the major part (59.13 to 61.07%). Another acid that stands out is oleic acid (17.45%). Other studies have also reported that these were the main fatty acids found in passion fruit seeds (Pereira et al., 2018; Silva and Jorge, 2017).

As can be seen in Table 3, stigmasterol and δ-tocopherol are the predominant phytosterols and tocopherols, respectively, in the composition of passion fruit seed oil.

In the extraction of tocopherols, the evaluated methods provided similar levels of these compounds. The values obtained (33.94 to 44.64 mg per 100 g of oil) were higher than those reported by Pereira et al., (2017) in the extraction using pressurized propane (9.87 to 11.14 mg per 100 g of oil) and UAE with ethanol (10.10 mg per 100 g of oil). However, they were lower than the value of 49.9 mg per 100 g of oil as reported by Malacrida and Jorge (2012) with the traditional extraction using petroleum ether.

The UAE provided an extract with a higher content in phytosterols of 166.64 mg per 100 g of oil. The content obtained in this study was lower than that obtained by Piombo et al., (2006) and Silva and Jorge (2017), who reported values of 209.74 and 274.67 mg per 100 g of oil, respectively.

The differences found in relation to the literature are related to the nature of the raw material used, as well as the solvent extractor and the efficiency of the extraction method used.

Table 4 shows the results of the chemical composition of the samples obtained after conducting the reaction step in 12 and 24 h. The analysis of the data must be carried out by observing which compound is of most interest in the product, which must be directed to the separation step aiming at the concentration of the target compound. It should be noted that in addition to the compounds identified (Table 4), these samples contain phospholipids, waxes and sugars (Baümler et al., 2016), compounds identified in vegetable oils obtained using ethanol as a solvent.

From the data presented in Table 4, it appears that in 12 h of reaction the product obtained contains ethyl esters, diglycerides (DG) and monoglycerides (MG) in higher concentrations. The high triglyceride conversion in the reaction transesterification products after 24 h of reaction was observed, obtaining a sample with a predominance in ethyl esters and MG under these conditions.

Ethyl esters can be directed to use as biofuels (Gonzalez et al., 2013; Stevanato and Silva, 2019), in compliance with current legislation. Bitonto et al., (2019) reported that these compounds are considered non-hazardous organic compounds, in addition to biofuels, and have industrial applications such as solvent, fragrances and cosmetics.

MG and DG have applications as emulsifiers in the food and pharmaceutical industry (Hartel et al., 2018). These emulsifiers are frequently used in bakery products, frozen desserts, and sauces/dressings (Nicholson et al., 2019). According to Ferreti et al., (2018) diglycerides are widely used for foods such as mayonnaise, salad dressings and in the confectionery industry, as they combine a hydrophilic head and a hydrophobic tail in the molecule that help hydrophilic and lipophilic substances to mix.

Thus, it was found that the sequential process provided a high-quality oil, eliminating the solvent removal step, which favors its application in different products and because it is a food residue, the oil obtained from the seeds Passion fruit can be applied to food products that require emulsifiers during their production process.