Palm kernel cake, pressed fibre, kernel shells and empty fruit bunches, which constitute by-product streams generated after the palm oil extraction process, were supplied by the company Companhia Refinadora da Amazônia (Belem, Brazil).

Methanol (HPLC grade), DPPH, ethanol (analytical grade), butylated hydroxyanisole (BHA) and standard compounds, such as pyrogallol, gallic acid, catechol, homovanillic alcohol, catechin, vanillin, lariciresinol, 4-hydroxybenzoic acid, myricetin, ferulic acid, syringaldehyde, guaiacol and sinapinic acid were purchased from Sigma Aldrich, Merck (Steinheim, Germany). Folin-Ciocalteu phenol reagent (2 M) was purchased from Merck (Darmstadt, Germany).

The extraction procedure was carried out in an ultrasonic bath (Elmasonic S10H; Elma, Long Island, NY, USA) using 5 g sample (on dry basis) and an aqueous solution of ethanol (70%, by volume) acidified with hydrochloric acid (0.01%, by volume) as a solvent. Preliminary extraction experiments were conducted to evaluate total phenolic content and antioxidant activity of the side streams. For this purpose, the extraction parameters were fixed as follows: 20 min extraction time and 30:1 liquid-to-solid ratio.

Subsequently, three different extraction times (10, 20 and 40 min) and four different liquid-to-solid ratios (5:1, 20:1, 30:1 and 40:1) were studied to identify the best conditions for the extraction of kernel cake phenolics. After extraction, the samples were filtered and extracts were evaporated at 40 °C under reduced pressure (rotary evaporator R-114; Büchi Labortechnik, AG, Flawil, Switzerland). The residue was redissolved in 50 mL acetone/water (1:1, by volume), washed four times with an equal volume of hexane, followed by ethyl acetate extraction (50 mL, four times). The ethyl acetate extracts were combined, and the solvent was evaporated under reduced pressure (19). The residue was redissolved in 10 mL methanol and kept at –20 °C until further analysis.

The total phenolic content of the methanolic extracts was determined using the Folin-Ciocalteu method, as previously described by Faustino et al. (20) and Papageorgiou et al. (21). Briefly, 50 μL extract, 450 μL distilled water and 2.5 mL 0.2 M Folin-Ciocalteu reagent were transferred into a test tube, vortexed (VTX-3000L Mixer Uzusio; LMS Co., Tokyo, Japan) for 1 min and allowed to stand for 8 min. Subsequently, 2 mL of an aqueous solution of Na₂CO₃ (75 g/L) were added and the mixture was left to react in the dark for 90 min at room temperature. The absorbance (relative to that of a blank prepared using methanol instead of extract) was measured at 765 nm using a double-beam UV-Vis spectrophotometer (model V-530; Jasco, Tokyo, Japan) and compared to the gallic acid calibration curve. The results were expressed in mg of gallic acid equivalents (GAE) per g of dry byproduct stream.

The antioxidant activity of the methanolic extracts was assayed using the DPPH free radical scavenging method as proposed by Scherer and Godoy (22). Briefly, 100 μL of diluted (1:2, 1:5, 1:10, 1:20 V/V) methanolic extract were added to 3.9 mL DPPH solution (80.1 μM in methanol) and left in the dark at room temperature for 90 min. The absorbance was measured at 517 nm against methanol as a blank, using a double-beam UV-Vis spectrophotometer (model V-530; Jasco). A mixture consisting of 100 μL methanol and 3.9 mL DPPH solution was used as a control. The antioxidant activity index (AAI) was calculated as suggested and described thoroughly by Scherer and Godoy (22):where IC₅₀ is the concentration of the methanolic extract providing 50% inhibition.

Chromatographic analysis was carried out on a Prominence HPLC system (Shimadzu, Kyoto, Japan), equipped with an on-line degassing unit (DGU-20A), a quaternary pump (LC--20A), an auto sampler (SIL-20A), a column oven (CTO-20A) and a photodiode array detector (SPD-M20A). Separation was performed on a Waters Nova-Pack C18 column (150 mm×3.9 mm, 4 μm particle size; Milford, CT, USA) at ambient temperature (25 °C) with gradient elution (solution A: 0.1% by volume formic acid in 5% methanol (Merck) and solution B: 0.1% by volume formic acid (Merck) in 100% methanol). The gradient programme was the following: 100% A at 0 min, 75% A for 15 min, 65% A for 40 min, 55% A for 60 min, 50% A for 65 min, 0% A for 90–95 min and 100% A for 110–120 min. The flow rate of the mobile phase was 1 mL/min and the injection volume was 20 μL (23).

Identification of the individual phenolic compounds was based on the comparison of the retention times and the UV spectra of unknown peaks to those of standard compounds. Quantitative analysis was based on calibration curves constructed at specific wavelengths of reference compounds using the external standard method.

A commercial sunflower oil (Athens, Greece) was supplemented with appropriate amounts of palm kernel cake and palm pressed fibre phenolic extracts. A known volume of methanolic extract isolated as described above was transferred into a test tube and the solvent was evaporated under a nitrogen stream. The residue was redissolved by the addition of 0.8 mL propylene glycol. Subsequently, 5 g of sunflower oil were added and the mixture was shaken vigorously (VTX-3000L Mixer Uzusio; LMS Co.) at 150 rpm, for 15 min at room temperature. Propylene glycol was employed as an aid for the dissolution of phenolics in the oily phase (19). The synthetic antioxidant BHA was also used for comparison at a mass per volume ratio of 0.02% (m/V), which is the maximum allowed.

A Metrohm Rancimat model 679 (Herisau, Switzerland) was used to measure the oxidative stability index before and after the enrichment of the sunflower oil with the palm kernel cake and pressed fibre extracts. The tests were carried out with 3 g oil samples at 120 °C and at an airflow rate of 20 L/h. Oxidative stability index was expressed as induction time (h).

JMP v. 8 software (24) was used for statistical analysis. The data were compared using analysis of variance (ANOVA) and Pearson’s linear correlation at 5% significance level. Significant differences between mean values were determined by honestly significant difference (Tukey’s HSD test) at p<0.05 and p<0.01. All values were reported as the mean of three independent determinations and the results were expressed as mean value±standard deviation.

Preliminary experiments were conducted in the four streams of oil palm solid wastes in order to select the most efficient one in terms of total phenolic content (TPC) and antioxidant activity index (AAI). Table 1 shows the TPC and the AAI of palm kernel cake, pressed fibre, kernel shell and empty fruit bunch extracts from two consecutive extraction cycles. ANOVA results show a statistically significant difference of the mass fractions of total phenolics among the above-mentioned waste streams (p<0.01). The comparison using Tukey’s test (Table 1) between total phenolic content of pressed cake and kernel shells did not show any statistically significant differences, with respective values (in mg GAE per g dry sample) of 3.59 and 3.66. Empty fruit bunches showed the lowest TPC (1.79 mg/g), while kernel cake exhibited the highest TPC, equal to 5.19 mg/g. Statistically significant differences were demonstrated between the total phenolic content of kernel cake and pressed fibre or empty fruit bunches or kernel shells, of pressed fibre and empty fruit bunches and finally of empty fruit bunches and kernel shells. The kernel cake extract had a relatively higher TPC than the ethanolic extract of sunflower meal, a byproduct from the biodiesel industry that had a TPC value of 4.4 mg/g (25) and the sesame cake extract that exhibited a TPC value of 0.81 mg/g (26).

We selected ultrasound-assisted extraction since various studies had proven that it leads to higher extraction efficiency than simple extraction by agitation (27). The negative pressure developed in an ultrasonic bath causes an alteration of the physicochemical properties of the plant matrix, leading to the formation of cavitation bubbles (28). As a result of this phenomenon, the cell membranes get disrupted, increasing the contact between the solvent and the matrix.

The extracts of the four side streams showed antioxidant activity. As estimated by the DPPH assay, the AAI of the extracts followed the order: empty fruit bunch<palm pressed fibre<kernel cake<kernel shells, varying between 0.18 and 1.48. Scherer and Godoy (22) proposed a scale classifying the extracts according to their antioxidant activity as poor if AAI<0.5, as moderate when 0.5<AAI<1.0, as strong when 1.0<AAI<2.0 and finally as very strong when AAI>2. Consequently, the extracts from empty fruit bunches are categorized as poor antioxidants, those of pressed fibre as moderate ones, whereas the extracts of kernel cake and kernel shells are considered strong antioxidants. A significantly positive correlation was obtained between TPC and AAI at R²=0.80 in all four waste streams. Velioglu et al. (29) obtained similar data reporting the TPC and antioxidant activity of 28 plants, including several fruits, vegetables, medicinal plants, as well as sunflower seeds and hull, flaxseeds and wheat germ. They demonstrated a significant relationship between these two factors in all plant materials except for the anthocyanin-rich materials and medicinal plants.

As the palm kernel cake extract had the highest TPC as well as strong antioxidant activity, it was selected for further investigation. Moreover, it is one of the main side streams of the palm fruit processing industry.

The extraction time and liquid-to-solid ratio were investigated in order to establish the best values of TPC and AAI of palm kernel cake extracts. Three different extraction times were tested (10, 20 and 40 min) and the obtained data are presented in Table 2. ANOVA analysis showed a statistically significant effect of the extraction time on the TPC of kernel cake (p<0.01). It was observed that the lowest TPC (2.78 mg/g) was obtained in the shortest extraction time (10 min), and the highest TPC (6.08 mg/g) in 40 min. There were statistically significant differences in TPC of kernel cake obtained at the three different extraction times at a liquid-to-solid ratio of 30:1. TPC was 15% higher after 40 min than after 20 min of extraction. Longer extraction times tend to result in an increased extraction efficiency. However, total phenolic potential of the solid matrix must be considered as well as the fact that extended extraction times could also affect the antioxidant activity (Table 2). Finally, Pearson’s linear correlation revealed a significantly negative correlation between the TPC and AAI (R²=–0.77). AAI was not significantly affected by any applied regime.

Subsequently, the liquid-to-solid ratios of 5:1, 20:1, 30:1 and 40:1 were investigated and their effect on TPC and AAI is given in Table 3. Increase in the liquid-to-solid ratio improved both TPC and AAI, revealing statistically significant differences as far as TPC was concerned (p<0.01). These results could be attributed to mass transfer phenomena and more specifically to the higher penetration of solvent into plant cells, leading to the release of more phenolic compounds and their easier diffusion into the solvent (30). Tukey’s test demonstrated that solely the ratio of 5:1 differs significantly from the other ratios (20:1, 30:1 or 40:1). Finally, a significantly positive correlation between TPC and AAI was demonstrated (R²=0.99).

These data are in accordance with other studies, for example Wong et al. (31) studied pH (2–6), temperature (30–80 °C), liquid-to-solid ratio (10:1–100:1), ethanol volume fraction (20–80%) and extraction time (30–300 min) during the extraction of phenolic compounds from palm kernel cake. Among these factors, the most important one was the liquid-to-solid ratio, which significantly affected both TPC and antioxidant activity. It was demonstrated that TPC and antioxidant activity increased when the liquid-to-solid ratio was increased from 10:1 to 100:1. However, the effect of extraction time on TPC content was not statistically significant, which is in contrast to our results. Additionally, Bucić-Kojić et al. (32) reported a significant difference in the yield of extracted polyphenols from grape seeds depending on liquid-to-solid ratio, with the highest polyphenol concentration obtained using a ratio of 40:1. Furthermore, Prasad et al. (33) applied a factorial design approach to identify the significant factors contributing to high extraction yield, antioxidant capacity and phenolic content in the extracts from Mangifera pajang pericarp. Ethanol concentration and liquid-to-solid ratio were reported as highly significant contributors.

Another important parameter that affects the extraction is the type of solvent. In the present study, ethanol was selected as the most suitable solvent for phenolic extraction because it is considered a green solvent and the ethanol extracts can be further used in food industry applications. The ethanol volume fraction applied was 70%. According to Markham (34), the extraction efficiency is higher when the solvent consists of a mixture of alcohol and water. Water causes the swelling of biomass and increases the contact surface area, resulting in improved extraction efficiency (35).

The main phenolic compounds derived from the extraction of the different palm oil wastes were identified by HPLC analysis (Table 4). Eleven compounds were identified in the kernel cake extract with pyrogallol being the predominant one (1550 μg/g), followed by 4-hydroxybenzoic acid (980 μg/g), gallic acid (590 μg/g) and ferulic acid (560 μg/g). Catechol, homovanillyl alcohol and catechin were also detected in significant amounts. In the empty fruit bunches, five compounds were identified with 4-hydroxybenzoic acid being the predominant one (550 μg/g). Four compounds were identified in the palm pressed fibre and five in the kernel shells. The main phenolic compound in the pressed fibre extract was again 4-hydroxybenzoic acid (760 μg/g), whereas pyrogallol was the predominant phenolic compound in the kernel shell extract (370 μg/g). Catechin and 4-hydroxybenzoic acid were present in all residues.

These results are in accordance with the ones obtained in the study of Neo et al. (13), who studied the phenolic content of extracts from oil palm fruits. In their study, the main phenolic compounds identified were 4-hydroxybenzoic, p-coumaric and ferulic acids in mass fractions 376, 55 and 114 μg/g, respectively. Ferulic and 4-hydroxybenzoic acids were also detected in various genera of the Arecaceae family (36). The authors suggested that 4-hydroxybenzoic acid could also be considered as a possible taxonomy marker for palms.

4-Hydroxybenzoic acid has already been applied as a preservative in pharmaceuticals, foods and cosmetics (36). The conventional synthesis of 4-hydroxybenzoic acid involves the chemical reaction of phenol, potassium hydroxide and carbon dioxide under high-pressure conditions. This process bears a high environmental impact not only due to the raw materials but also due to the production of toxic liquid wastes. Finding renewable sources of this phenolic compound is an interesting alternative towards a more sustainable economy. Kawamura et al. (37) proposed that low-molecular-mass phenolic compounds could be used for the production of 2-pyrone-4,6-dicarboxylic acid, which is a precursor for the production of various polymers such as polyamide, polyester and polyurethane.

Pyrogallol and catechol are synthesized in plants through the shikimate pathway. Catechol has been found in extracts from the leaves and needles of various deciduous and coniferous trees, whereas pyrogallol is considered as a product of the degradation of plant debris. Moreover, these phenolic compounds are typically found in the soil, and they are also precursors of humic-like polymer synthesis (38). Catechin is a well-known flavonoid and is a plant secondary metabolite. Palm oil is rich in catechin, which has also been found in methanolic leaf extracts of E. guineensis together with epicatechin, epigallocatechin, quercetin and rutin among others (39).

Various residues of oilseed crops have been studied for the extraction of phenolic compounds such as sunflower meal and rapeseed meal (25, 40). The presence of gallic, vanillic, 4-hydroxybenzoic, ferulic, caffeic and syringic acids is very common in oilseed crops.

Further fractionation of this crude phenolic extract could increase its application and at the same time its market potential. Simple phenolics such as gallic acid, catechol and pyrogallol could be employed to produce fine chemicals, resins, pharmaceuticals and also for food processing and preservation (41). Fini et al. (42) attempted to use pyrolytic lignin instead of fossil bitumen to produce phenolic resins.

Pyrogallol, which is the main phenolic compound identified in the extracts, has a broad range of applications, such as in food, cosmetics, chemical and pharmaceutical industries as well as in photography, printing and production of dyes, due to its antioxidant and antibacterial properties (43). A pyrogallol-based oxygen scavenging system with sodium carbonate was proven to be effective for food packaging applications with low water activity (44). Cotton fabric was also treated with pyrogallol, leading to a final product with >99.9% antibacterial ability and >90% antioxidant activity (45). Depending on the application purposes and its purity, the price of pyrogallol varies from $20–100/kg, depending on the supplier.

Ferulic acid, with a broad range of applications, was also present in the extract. This phenolic compound has been reported to increase shelf life of linseed oil and to enhance the stability of lard and soybean oil (46). A mixture of ferulic acid and glycine was proven to completely inhibit the oxidation of biscuits (46). Ferulic acid is also a precursor of vanillin, a well-known flavouring agent widely used in food, beverages and cosmetics (46). Food- or pharmaceutical-grade ferulic acid can cost from US $100–1000/kg, again depending on its purity, origin and the supplier.

The Rancimat method is widely used to determine the oxidative stability of edible oil. The phenolic extracts from palm kernel cake and pressed fibre were used to enrich sunflower oil in order to increase its stability against oxidation. Palm kernel cake extracts were selected since they exhibited the highest TPC value, and the highest total phenolics as measured by HPLC. Pressed fibre extracts were also tested, mainly for comparison, since it was the stream that, after kernel cake, had the second highest total phenolic mass fraction, determined with HPLC. In both cases, the entire extract was redissolved in propylene glycol and added to 5 g of sunflower oil. After extraction, the total dry mass of the kernel cake crude extract was 42 mg, while the corresponding value of the pressed fibre was 23.2 mg. These values correspond to an addition of 0.8% of kernel cake and 0.46% of pressed fibre extract.

Butylated hydroxyanisole (BHA) is a synthetic antioxidant commonly used in the food industry because it is cheaper than natural antioxidants and at the same time very efficient (26). For comparison, BHA was also tested at its legal limit (0.02%). Sunflower oil without the addition of any antioxidant was also used as control. The results of the Rancimat test are shown in Fig. 2. According to ANOVA, the effect of the extract type on the induction time of the sunflower oil was statistically significant (p<0.01). Tukey’s test revealed statistically significant differences among the tested samples (Fig. 2) except for the comparison between BHA and 0.8% palm kernel cake extract, which did not show any significant differences. The induction time of the control was only 2.6 h. The addition of 0.46% palm pressed fibre extract slightly increased the oil induction time (3.4 h). The pressed fibre extract increased induction time up to 30%, and kernel cake extract almost 60%, a result very similar to the one obtained with BHA. The results are very promising, especially considering that BHA is a pure compound, while the palm kernel cake extract also contains other substances of non-phenolic nature that could even act as pro-oxidants.

In the literature, there are several studies regarding the addition of extracts from various plant sources to sunflower oil. Gamel and Kiritsakis (47) reported that a 0.02% rosemary methanolic extract increased the oxidative stability of sunflower oil at 63 and 120 °C. Yanishlieva et al. (48) established that the ethanol extract of summer savory (0.1–0.5%) improved the oxidative stability of sunflower oil after 50 h at 180 °C. Finally, the antioxidant activities of six plant extracts (catnip, hyssop, lemon balm, oregano, sage and thyme) added to sunflower oil were evaluated. Sage extracts (600 and 1200 mg/L) effectively inhibited the oxidation process in sunflower oil and had higher antioxidant activity than BHT 300 mg/L (49).