The passion fruit species (Passiflora alata BRS Doce Mel, Passiflora setacea BRS Peróla do Cerrado and Passiflora tenuifila) were cultivated in the experimental Embrapa Cerrados fields, located at Planaltina-DF/Brazil. The fruit was harvested when the skin color of the fruit reached about 20% yellow (between 40 and 60 days after inflorescence). The assays were carried out using homogenized seed lots between January and May, 2016. After fruit pulping, all the Passiflora seeds were stored at −20 °C until use. After thawing, the seeds were washed under tap water to remove the pulp and mucilage residues adhered to the seeds, followed by autoclaving at 120 °C for 20 minutes in order to reduce microbial loads.

Passion fruit seeds were dried in a convective dryer (Hauber – DMS-P) at 50 °C and 2-g samples were weighed at 30 minute intervals until constant weight over three successive weights. The drying temperature was optimized in previous experiments for P. alata and P. setacea seed drying (Paula, 2015). Physical air properties were monitored by the dew point of a Cole-Parmer Traceable® hygrometric thermometer. The room temperature of the air was 22.7 ± 1 °C and humidity was approximately 50% and the temperature and humidity inside the dryer were 45 ± 1 °C and 15%, respectively. Drying was carried out in triplicate.

In order to estimate the effective water diffusivity (D_eff) during seed drying, the dimensionless moisture (MR) and time (t) data were adjusted by applying the second Fick diffusion law model (Eq. 1) in spherical coordinates, considering up to three term models of the infinite series solution, using the Statistica software (v13.0).

After drying, P. alata and P. setacea seeds were ground using a knife mill (Macron MA-048). This step was not necessary for P. tenuifila seeds, as they displayed a reduced mean diameter (dp < 3 mm). The seeds’ moisture was adjusted between 12 and 15% in sealed dissectors containing distilled water, as recommended by Evangelista and Cermak, 2007. Crushed passion fruit seeds were processed in a continuous press (Oekotec - Germany, model CA59G), at room temperature of 22 ± 1 °C and compression ratio of 4.4 to obtain the first-extraction crude oil, which was then decanted for 24 h in a dark cabinet. A micropipette (Labmate Soft 1–5 μL) was used to separate the clarified oil. The clarified fraction was stored at −18 °C until analysis. To improve the extraction efficiency, the pressed cake was recycled and reprocessed at 30 ± 2 °C. At the end of extraction, the residual oil in pressed cake was recovered by Soxhlet using petroleun ether as solvent and the efficiency was then calculated. The defatted pressed cake was reserved for subsequently extraction of phenolic compounds.

For phenolic compound extraction, the pressed cake was ground using a knife mill (Macron MA-048) coupled to a 1 mm diameter circular mesh sieve. The crushed sample, with particle diameter of less than 2 mm, was then homogenized with ethanol at a 1:5 ratio (pressed cake:ethanolic solution).

The experimental design was carried out with three different aqueous-ethanol solutions selected from the preliminary test at the best selected extraction parameters of 30:70, 50:50 and 70:30 ethanol:water. The suspension was maintained in a heated bath at 45 °C for 30 minutes at a stirring rate of 600 rpm. After extraction, the solids and soluble compounds were separated by vacuum filtration. The liquid phase containing the bioactive compounds was stored at –18.0 ± 0.5 °C until analysis.

Oxidative Stability: The oxidative stability of passion seed oils was determined using Rancimat® 743 (Metrohm, model 743) equipment, according to EN 14112, under a 10 L·h⁻¹ flow at 110 °C, using 3 g of each oil sample, in duplicate. These test data were expressed as an induction period that represented the time interval until the sample reached a high oxidation level.

Antioxidant Capacity: The antioxidant capacity of passion seeds oils was determined according to a modified methodology developed by the Embrapa Tropical Agroindustry for fruit analysis (Rufino, 2007), which uses the DPPH (2,2-diphenyl-1-picryl-hidrazil) radical method, adapted for vegetable oils. The DPPH reagent was added at five different dilutions of the original oil in isopropyl alcohol, which then remained for one hour in the dark prior to a spectrophotometric analysis at 515 nm ((BelFotonix SP 1105). The Inhibitory Concentration (IC50 - mass necessary to reduce the initial concentration of the DPPH radical by 50%) was determined from the absorbance readings.

The antioxidant capacity of the pressed cake was evaluated by the ABTS method, expressed as Trolox equivalents (TEAC) according to Re et al., (1999) and Rufino et al., (2007). The ABTS radical was chemically obtained by ABTS oxidation by potassium persulfate. Aliquots of 5 mL of the aqueous 7 mM ABTS solutions were mixed with 88 μL of 140 mM potassium persulfate in amber bottles. The flasks were kept at room temperature for 14 h for complete reaction and radical stabilization, and a concentrated radical solution was obtained. The concentrated ABTS + solution was diluted in 95% ethanol to an absorbance of 0.700 ± 0.020 at 734 nm (BelFotonix SP 1105) to form the test reagent.

Trolox® (300–2000 μM in 95% ethanol) and the extracts were transferred to test tubes where they were left to react for 6 minutes with the diluted ABTS + solution at a 1:10 ratio at room temperature, followed by immediate spectrophotometric absorbance determinations at 734 nm (BelFotonix SP 1105). The discoloration of the mixed solution indicated that the antioxidant compounds in the extract inhibited the radical ABTS cations. A quantitative relationship between absorbance reduction at 734 nm and antioxidant concentrations present in the sample was noted. The results were expressed as μmol Trolox·g⁻¹ of extract Total Phenolic (TP) Content: The total phenolic (TC) content of the pressed cake extract was determined by the spectrophotometric method proposed by Singleton and Rossi (1965), using the Folin-Ciocalteu reagent. This method is based on the oxi-reduction reaction, which forms a blue compound with the phenol. Briefly, 0.5 mL of each sample was mixed with 2.5 mL of the 10% Folin-Ciocalteu reagent and vortexed for 2 min. After homogenization, 2 mL of a saturated sodium carbonate solution (7.5%) were added. The solution was then incubated for 15 min at 50 °C in a water bath, and immediately cooled in an ice bath for 30 seconds. The absorbance of the mixture was determined at 760 nm using a spectrophotometer (BelFotonix SP 1105). Phenolic compound quantification was performed using a gallic acid calibration curve. The results were expressed as mg gallic acid equivalent per 100 g of sample (GAE.100.g⁻¹).

The extracts obtained from the passion fruit seeds P. alata, P. setacea and P. tenuifila in 70:30 ethanol-water solutions were selected to increase the oxidative stability of their respective oils. To maintain bioactive properties, the solvents were removed from the extracts by forced convection at an air temperature of 21 °C.

The dried extracts were mixed with passion fruit seed oils at 10% mass/mass ratio, according to preliminary tests. To evaluate the oxidative stability of the mixtures, 3 g of each sample were then transferred to a Rancimat reaction tube, and the analyses were performed in duplicate using an air flow at 10 L.h⁻¹ and 110 °C.

All analytical determinations were carried out at least in triplicate, except for oxidative stability trials. An analysis of variance (ANOVA) followed by Fisher’s LSD test was performed using the Statistica® v.13.0 software.

The estimated drying rate parameters and corresponding R-squared results are presented in Table 1. In addition, kinetic curves at 50 °C which fitted the experimental data for passion fruit seed drying by Fick’s second law are displayed in Figure 1.

The following equations were used to determine P. alata (Eq. 1.a), P. setacea (Eq1.b) and P. tenuifila (Eq. 1.c):

The diffusion coefficient for P. alata seeds was higher than the other two assessed species. It was possible to estimate the drying time of the passion fruit seeds from the fitted equation, in order to achieve the best moisture for pressing passion fruits seeds between 10 and 15%.

The oil contents of dried passion fruit seeds were 19, 16 and 23% for P. alata, P. setacea and P. tenuifila, respectively (Table 2). Paula (2015) reported a lipid content of 22.5 and 32.2% for P. alata and P. setacea, respectively. This difference may be due to seasonality effects, particle size and moisture. The efficiencies of the continuous oil extraction process, calculated by the ratio of the oil mass obtained for the total oil in the seeds, were 57% to 67%, as displayed in Table 2.

These results are typical of continuous oil extraction from lignin-rich seeds (up to 40% in dry matter and lipid content of less than 30%). In addition, the extraction yield for P. setacea was lower, in contrast to the data reported by Paula (2015), possibly due to the harvest season. The overall yield after heated cake pressing was increased by about 10%. According to reported data, cold-pressed oils require no further refinement and can be consumed after simple decantation, presenting superior sensory characteristics and nutritional qualities when compared to hot-pressed oils (Emir, Gunes and Yılmaza, 2014).

As displayed in Table 3, the induction periods (IP) for P. setacea and P. tenuifila seed oils at 110 ºC were 7.32 h and 6.87 h, respectively, which were typical IP for passion fruit seed oils (Paula et al., 2015). However, P. alata presented a lower IP 50% (3.53 h), which could be explained by the lower oleic acid and higher linoleic acid contents in P. alata seed oil (Santana et al., 2015). Due to the two double bonds present in linoleic acid, this acid is 40 times more unstable than oleic acid, which contains only one double bond (Damodaran and Parkin, 2017).

The IC₅₀, defined as the amount of antioxidant required to reduce DPPH concentrations by 50%, is considered a good antioxidant capacity measure (inversely proportional). Regarding DPPH results, the oil from P. setacea seeds presented the highest antioxidant capacity, followed by P. alata and P. tenuifila. These values were consistent with the literature data reported by Santana et al., (2015), and could be explained due to the higher vitamin E content in P. setacea (215.3 mg·100 g⁻¹) and P. alata (114.8 mg·100 g⁻¹) seed oil compared to P. tenuifila seed oil (80 mg·100 g⁻¹).

The most efficient solid-liquid extraction of bioactive compounds from the pressed cake was achieved using 70:30 ethanol-water solutions at 45 °C. This was probably due to the greater solubility of the polar compounds in the passion fruit seed cake in this aqueous and ethanolic solution. The highest values for total TP contents and TEAC for the pressed cake extracts are displayed in Table 4. For the selected experimental conditions, the ethanolic P. setacea extract presented higher TP contents and TEAC when compared to the other species.

Total phenols from P. setacea, P. alata, and P. tenuifila cakes were approximately 1800, 600 and 900 mg GAE·100 g⁻¹ of extract, respectively. The TP of the extract obtained at a 70:30 ratio from the P. setacea cake was 3- and 2-fold higher than the P. alata and P. tenuifila extracts, respectively (Table 4). A strong positive Pearson correlation (R=0.97) with antioxidant activity was observed, since the results for the antioxidant capacity in the same experimental conditions for the P. setacea cake were 2.3- and 1.8-fold higher than for P. alata and P. tenuifila, respectively.

According to Vasco et al., (2008) the ethanolic extracts obtained in the present study from pressed P. alata and P. tenuifila cakes can be treated as medium phenolic contents, while the extracts obtained at 50:50 and 70:30 ethanol:water ratios from pressed P. setacea cake are treated as high phenolic contents.