Hexane, acetonitrile, trimethylamine, methanol, ethyl acetate, isopropanol and tetrahydrofuran, all HPLC grade, were obtained from Dinâmica (Diadema, Brazil), J. T. Baker (Phillipsburg, USA) or Vetec (Rio de Janeiro, Brazil). The samples and solvents were filtered through Millipore (Billerica, USA) membranes (0.22 and 0.45 µm) prior to HPLC analysis. Sodium hydroxide, potassium hydroxide, methanol, petroleum ether and ethyl ether were obtained from Labsynth (Diadema, Brazil) or Dinâmica (Diadema, Brazil) and were of analytical grade. Standards of β-carotene and tocopherols were obtained from Sigma-Aldrich (St. Louis, USA).

Six liters of crude buriti oil (Amazon Oil^TM) were obtained in Ananindeua city, Pará state, in the Brazilian Amazonian region and stored at -30 °C until analysis.

The characterization of buriti oil was carried out by physicochemical analyses, thermophysical properties, β-carotene and tocopherols contents, fatty acids and acylglycerol classes and thermal analyses (crystallization and melting).

The physicochemical characteristics of buriti oil were assessed by the determination of free acidity (expressed in equivalent of oleic acid) and peroxide, iodine, saponification and refractive indexes, according to AOCS (2009). All the analyses were carried out in triplicate.

Rheology and density of crude buriti oil were determined in triplicate at 25 ºC.

Rheology. A controlled stress rheometer, model ARG2 (TA Instruments, New Castle, USA), was used to carry out steady state flows with cone-plate geometry (52 μm gap). The shear rate varied from 0.1 s^-1 to 500 s^-1, and the mean shear stress was acquired through the data acquisition system. Then, the resulting rheograms were fitted to Newton model to provide the viscosity value.

Density. The density was determined using a digital electronic density meter (model DMA 4500-M, Anton Paar, Austria), in which approximately 10 mL of crude buriti oil were used for each run. After establishing the temperature, the density values were provided by the equipment.

Carotenoids. The content of β-carotene in buriti oil was determined by high performance liquid chromatography coupled to a diode array detector (HPLC-DAD). Before the injection into the HPLC system, the crude buriti oil was submitted to saponification in 10% KOH in methanol for 14-16 hours, followed by a liquid-liquid partition using petroleum ether:ethyl ether (1:1, v/v) and the alkali was removed by washing with distilled water (De Rosso and Mercadante, 2007). The carotenoid extract was injected into a Shimadzu HPLC (Kyoto, Japan) equipped with quaternary pump (LC-20AT) with a 10 µL loop, on-line degasser, auto sampler (SIL-20A) and a DAD detector (SPD-M20A). The compounds were separated on a C18 Shim-pack VP-ODS column (250 mm x 4.6 µm) using, as mobile phase, a linear gradient of acetonitrile (0.05% triethylamine):methanol:ethyl acetate from 60:20:20 to 60:0:40 in 43 minutes and maintaining this proportion for 17 min. The flow rate was 1 mL/min, the column temperature was set at 29 °C and the chromatograms were processed at 450 nm (Silva et al., 2014). β-carotene was identified according to the following combined information: elution order on C18 column, co-chromatography with authentic standard, and UV–Vis spectrum [λ_max, spectral fine structure (%III/II), peak cis intensity (%AB/AII)] compared with the data available in the literature (De Rosso and Mercadante, 2007). The β-carotene was quantified (in triplicate) by HPLC-DAD, using external seven-point analytical curves (in duplicate) with all-trans-β-carotene standard (0.547–35.0 µg/mL). Total carotenoids were considered as the sum of the carotenoid peak areas in the chromatogram obtained.

Tocopherols. The tocopherol profile and total tocopherols in buriti oil were determined according to AOCS Ce 8-89 methodology (AOCS, 2009), using the high performance liquid chromatography coupled to fluorescence detector. Before the injection into the HPLC system, the crude buriti oil was solubilized in hexane (1% w/v). The samples were injected into a Perkin Elmer SERIES 2000 HPLC, under the following chromatographic conditions: isocratic pump Perkin Elmer 250; fluorescence detector Shimadzu RF-10 AXL with excitation-290 nm and emission-330 nm; Merck analytical column 250 x 4 mm Li Chrosorb Si 60*5μm coupled to compatible guard column; mobile phase: 99:1 hexane/isopropanol (filtered and degased for 10 minutes in an ultrasonic bath), flow 1.0 mL/min; injected volume of 20.0 μL. The individual tocopherols were identified by elution order on the analytical column compared with external authentic standards and quantified (in triplicate) using external analytical curves with authentic standards. Total tocopherols were considered as the sum of the individual tocopherol concentration.

The determination of the fatty acid composition was performed on a gas chromatograph with capillary column - CGC Agilent 6850 Series GC System, after esterification using the Hartman and Lago (1973) method. The fatty acid methyl esters were separated according to the AOCS Ce 1-62 method (AOCS, 2009) on capillary column DB-23 Agilent (50% cyanopropyl-methylpolysiloxane) with dimensions: 60 m, φ int: 0.25 mm, 0.25 μm film. Oven temperature 110 °C-5 min, 110-215 °C (5 °C/min), 215 °C-24 min; detector temperature 280 °C; injector temperature 250 °C; helium drag gas; split ratio 1:50; injected volume of 1.0 μL. The qualitative composition was determined by comparing peak retention times with those of respective fatty acid standards. Mustard oil was used as standard. The quantitative composition was determined by the peak percentage area.

The buriti oil acylglycerol classes were determined by high performance size exclusion chromatography (HPSEC) using liquid chromatograph Pelkin Elmer Series 200; refractive index detector Waters 2414; two columns, being 1 - JORDI GEL DVB 300 x 7.8 mm, 500 Å and 2 - JORDI GEL DVB 300 x 7.8 mm, 100 Å; mobile phase tetrahydrofuran; flow rate 1 mL/min; injected volume 20.0 μL. The sample was solubilized in tetrahydrofuran at 1.0% concentration. The acylglycerol separation mechanism by the HPSEC technique occurs through the molecular size difference among compounds present in the reaction enviroment. In this case, the elution order is triacylglycerols, diacylglycerols, monoacylglycerols and free fatty acids.

The buriti oil triacylglycerol composition was estimated using the software PrOleos, based on the hypothesis of 1,3-random-2-random distribution, which predicts triacylglycerols’ molar percentage present in vegetable oils, from the fatty acid composition (Antoniosi Filho et al., 1995). This software is available in the plataform “https://lames.quimica.ufg.br/p/4035-material-didatico”. The triacylglycerol classes were estimated by triacylglycerol composition.

Thermal analysis for buriti oil crystallization and melting behavior was performed by differential scanning calorimetry using DSC 8000 (Pyris Series, Perkin Elmer) at 80 °C for 10 min, 80 at -80 °C (5 °C/min), -80 °C for 10 min, -80 to 80 °C (5 °C/min).

The buriti oil used was crude, and due to this fact a higher percentage of free fatty acids was observed (Table 1) compared to refined oils. For example, Ribeiro et al. (2009) reported 0.03% FFA for refined soybean oil. The peroxide index is related to the oil oxidative state and not to its characterization itself. Santos et al. (2013) found 7.4 meq O₂/kg for buriti oil and reported this value within expected ranges for crude oils of good quality. The saponification and iodine values are related only to specific characteristics of each vegetable oil, such as chain length and unsaturation number (Pardauil et al., 2017). Both the iodine and saponification indexes found for buriti oil were similar to those found in the literature, being 73 g/100 g and 193 mg KOH/g, respectively (Lognay et al., 1987). The refractive index found for buriti oil (Table 1) was the same as previously reported in the literature (Albuquerque et al., 2005; Silva et al., 2009).

With respect to rheology, buriti oil showed Newtonian behavior as expected in the literature for oils (Steffe, 1996). It was characterized by the linear dependence between shear rate and shear stress with R²>0.9999 and null intercept. Newtonian viscosity presented close values to the ones published by Santos et al. (2005) for different vegetable oils. In the same way, the viscosity of buriti oil was comparable to the values found by Ribeiro et al. (2017) for different commercial olive oils. These olive oils had a very similar fatty acid composition to buriti oil, which also resulted in close agreement between buriti and olive oil densities. Besides being an indicative of degradation, such properties are useful for the correct design of oil pumping, settling and filtration (Bonnet et al., 2011).

Buriti oil presented high contents of carotenoids (686.89 mg/kg), with β-carotene as the major compound, accounting for 74% of total carotenoids (506.84 mg/kg), which imparts the oil its orange-red color. Godoy and Rodriguez-Amaya (1994) emphasize that buriti is a very important pro-vitamin A source, since its content in β-carotene is much higher than that found in most of Brazilian fruits. Crude palm oil shows total carotenoids of 608.39 mg/kg (Ferreira et al., 2016), which is similar to the crude buriti oil in our study. In addition, natural sources of this carotene are of great interest because only 2% of commercial β-carotene is naturally produced (Dufossé et al., 2005; Ribeiro et al., 2011; Speranza et al., 2016).

Buriti crude oil was reported with a high amount of total carotenoids by Lognay et al. (1987) (1730 mg/kg), Albuquerque et al. (2005) (1707 mg/kg) and Silva et al. (2009) (1003 mg/kg) with β-carotene present as the principal component.

De Rosso and Mercadante (2007) reported the identification and quantification of carotenoids in buriti pulp by HPLC-DAD using a C₃₀ column and coupled to a mass spectrometer (APCI ionization source), and presented β-carotene as the major carotenoid (72%), followed by 13-cis-β-carotene (12%), 9-cis-β-carotene (3.6%), γ-carotene (2.9%) and cis-γ-carotene (1.9%) and a total carotenoid content of 513.87 mg/kg. Santos et al. (2015) investigated the carotenoid composition in buriti oil by HPLC-DAD. The authors reported a total carotenoid content of 54.81 mg/kg with β-carotene as the principal carotenoid (55%), followed by cis-β-carotene (31%), lutein (6%) and cis-lutein (3%).

Although in our study other carotenoid peaks were observed in the chromatogram of the crude buriti oil sample, it was not possible to perform the identification by the HPLC-DAD analysis, since cis and trans isomer separation were not efficient and the cis peaks were present at very low intensity. In the chromatogram (Figure 1) it is possible to notice the carotene peaks and β-carotene peak with the biggest area.

Concerning the tocopherol profile, which can be seen in Table 2, buriti oil presented a content of total tocopherols of 104.14 mg/100 g. The principal tocopherol was β-tocopherol (48%) followed by α-tocopherol (43%), in this way, they were present in similar amounts and represented 91% of total tocopherols.

Lognay et al. (1987) and Albuquerque et al. (2005) reported similar values for total tocopherols, but Lognay et al. (1987) presented the α-tocopherol as the major component (48.1%) while (β + γ)-tocopherol represented 48.3%. Silva et al. (2009) and Santos et al. (2013) reported high values for total tocopherols at 152 mg/100g and 156.7 mg/100 g, respectively. While Silva et al. (2009) presented β-tocopherol as the major component (45%) followed by α-tocopherol (40%), Santos et al. (2013) presented α-tocopherol as the major component (70%) followed by β-tocopherol (30%). These results reinforce the fact that buriti oil has a promising potential as a dietary source due to its high vitamin E content (Santos et al., 2013). Such difference in reported values of unsaponified matter might be due to fact that the composition and, consequently, the nutritional value of crude buriti oil vary with the variability and/or ripening stage, the region of provenance and extraction process (Santos et al., 2015; Silva et al., 2009).

Buriti oil presented higher levels of unsaturated fatty acids than saturated fatty acids, being oleic acid the main fatty acid in buriti oil (69.58%) followed by palmitic acid (17.35%) (Table 3). Similar compositions were also related by other authors (Escriche et al. 1999; Albuquerque et al., 2005; Silva et al., 2009; Bataglion et al., 2015).

The profile of fatty acids found in buriti oil revealed a good source of monounsaturated fatty acids. According to Speranza et al. (2016), a great interest has been placed on oils that contain these fatty acids. High oleic and low linoleic fatty acid contents help to make them more resistant to oxidation than most oils.

From the point of view of nutritional and food industries, in general, oleic acid is the most abundant omega-9 fatty acid in food and one of the most commonly found fatty acids in nature. It is monounsaturated, considered fundamental for beneficial properties in the reduction of LDL-cholesterol oxidation, besides being a precursor for the production of most other polyunsaturated fatty acids and hormones (Syed, 2012; Watkins and German, 2008; Angelis, 2001).

Regarding the acylglycerol classes, the most abundant compounds are triacylglycerols (88%) (Table 4). Such result was similar to that reported in the literature; 92.6% of triaclylglycerols and the content of diacylglycerols, monoacylglycerols and free fatty acids were about 5% for buriti crude oil (Lognay et al., 1987).

The predominant triacylglycerols were OOO, POO, OLO and POP, representing 80.12% of the total (Table 5). It was observed that the composition of triacylglycerols in buriti oil was extremely homogeneous, considering the low fatty acid variability of this matrix, given the high contents in oleic and palmitic acids followed by linoleic acid. This fact also justifies the large percentage of triunsaturated (U₃) and diunsaturated (SU₂) triacylglycerols in the buriti oil composition of triacylglycerol classes (Table 6).

Santos et al. (2013) determined the buriti oil triacylglycerol profile by HPCL reporting results for OOO (39.8%), POO+PLS (35.9%) and POP (10.2%) and by GC reporting results for OOO (35.6%), POO (38.8%) and POP (9.4%). These similar results could confirm the good accuracy of the PrOleos software, in which, Speranza et al. (2016) obtained a similar triacylglycerol profile for buriti oil.

Antoniosi Filho (1995) and Lognay et al. (1987) also found that OOO, POO and POP were the predominant triacylglycerols in the buriti oil analized, but they did not observe OLO as one of the predominant ones, reporting that it corresponded to 1.1 and 1.8% of total triacylglycerols, respectively.

The crystallization temperature range was observed between -58.54 and 1.16 °C, with the peak at -43.06 °C. In turn, the melting temperature range was observed between -24.51 and 18.52 °C, with the peak at -2.73 °C. The crystallization and melting curves can be seen in Figure 2.

Because buriti oil is a mixture of triacylglycerols, crystallization and melting are observed in a temperature range, but not at a specific temperature. This is due to the fact that triunsaturated triacylglycerols present crystallization and melting points different from diunsaturated, monounsaturated and trisaturated triacylglycerols as a consequence of the different degrees of unsaturation.

It is interesting to note that the crystallization range did not coincide with the melting range, which is a common behavior in oils as a result of the different triacylglycerol arrangements when oil is liquid or solidified.

Buriti oil has approximately 90% of triunsaturated and diunsaturated triacylglycerols, and these have lower crystallization points than monounsaturated and trisaturated ones. When oil is liquid and the temperature is gradually reduced, monounsaturated and trisaturated triacylglycerols are melted in diunsaturated and triunsaturated and these last ones crystallize at a lower temperature. This behavior justifies a low crystallization range.

On the other hand, when buriti oil is solidified, triacylglycerols are arranged in a more organized way, which makes the melting range higher than the crystallization and melting transitions of highly polyunsaturated fatty acids followed by the melting of monounsaturated fatty acids and a small fraction of saturated fatty acids (Pardauil et al., 2017). Pardauil et al. (2017) reported a similar crystallization curve and according to these authors and Tan and Che Man (2000), crystallization curves are reproducible and simpler than melting curves due to the oil crystallization being influenced only by chemical composition and not by the initial crystalline state which is a consequence of the polymorphism phenomena of natural oils, and this cannot be established unequivocally by DSC measurements.

Crystallization and melting curves also depend on fatty acid composition. Usually, oil samples with a high degree of saturation show higher temperatures of crystallization and melting profiles than oils with a high degree of unsaturation (Pardauil et al., 2017).

The triunsaturated triacylglycerols, prevalent in buriti oil, present melting points from −14 to 1 °C and they are important for the softness and lubricity of products at room temperature. In addition, the diunsaturated triacylglycerols, with melting points from 1 to 23 °C, are important for the oral properties and mechanical-performance of products at room temperature (O’Brien, 2009; Rodrigues and Gioielli, 2003; Bessler and Orthoefer, 1983; Speranza et al., 2016).