The antioxidant activity compounds were evaluated under Rancimat based on an earlier method reported by Olajide et al., (2020) with some modification. 3 ± 0.05 g lard samples containing 200 ppm antioxidants were subjected to accelerated oxidation between 100 and 140 °C at an air flow rate of 20 L/h. The results were expressed as IP of antioxidant-spiked samples in hours relative to those without antioxidants. The protection factors (Pf) of antioxidants were also calculated with the formula below:

where IP_s is the induction period of oil with added antioxidant, and IP_b is that of oil without antioxidant.

The DPPH radical scavenging capacity was carried out according to the procedure described by Jiang et al., (2014) with slight modification. Briefly, 0.5 ml antioxidant in methanol (1.5, 3, 6, 12, 24, and 48 µM) was added to 2.5 ml DPPH methanolic solution (0.1 mM), vigorously mixed, left to react in a dark chamber for 30 min, and the decreasing absorbance of DPPH was read at 517 nm against a blank on a spectrophotometer. Methanol served as blank and 2.5 ml solution of DPPH plus 0.5 ml methanol were used as the control. EC₅₀, the effective concentration required to obtain 50% antioxidant capacity of compounds was calculated from the linear regression of plots between the scavenging activity (%) of antioxidants and their concentrations. DPPH radical scavenging activity was calculated with the formula below:

The peroxide and p-anisidine contents in oil samples were measured according to the AOCS Official Methods Cd 8b-90 (AOCS, 2011) and 18-90 (AOCS, 1995) with some modification.

Finally, Totox was calculated with the following equation:

The structure-antioxidant activity of the new antioxidants, 5a and 5b was evaluated in comparison with GA, BHT and TBHQ under temperatures of up to 140 °C at 0.02% (w/w) under air saturation conditions. The results are expressed as the IP of samples compared to the oxidative stability of lard without antioxidants (Table 1), and the Pf of the antioxidants (Figure 1). Pf values greater than 3 indicate strong antioxidant activity and those lower than 1 indicate pro-oxidant activity (Weng and Huang, 2014). In this experiment, the concentration was set at 200 ppm according to the safety limit established for frequently used and emerging commercial antioxidants in oils (Saad et al., 2007).

The IPs of GA, BHT and TBHQ were less than 5a and 5b at the different temperatures. i.e., the novel lipophilic GA derivatives exhibited much stronger oxidative stability than GA and the common commercial synthetic antioxidants, TBHQ and BHT. Compound 5b, however, showed the highest IP, which was 146 h higher than 5a and 159, 209, and 170 h more than GA, BHT and TBHQ, respectively. For all the antioxidants evaluated at 0.02% (w/w) (Figure 1), a Pf greater than 1 was observed, which affirms a protective capacity against accelerated oxidation in the lipid matrix. Nevertheless, 5b similarly exhibited the highest Pf (110 °C, Pf = 22.21; 140 °C, Pf = 18.07), followed by 5a (110 °C, Pf = 7.98; 140 °C, Pf = 9.78), GA (110 °C, Pf = 6.65; 140 °C, Pf = 6.19), TBHQ (110 °C, Pf = 5.71; 140 °C, Pf = 4.70) and finally, BHT (110 °C, Pf = 1.93; 140 °C, Pf = 2.37). Briefly, the antioxidant activity of the evaluated antioxidants under Rancimat conditions decreased as follows: 5b >> 5a > GA > TBHQ > BHT > Control.

Previous studies have shown GA and its alkyl esters to exert better antioxidant activities than free polyphenols at moderate temperatures (Farhoosh et al., 2016) and even TBHQ (Jung and Choi, 2016) in lipid matrices. Nonetheless, the much stronger antioxidant effects were exhibited by 5a and 5b compared to GA, The TBHQ and BHT in this study may be attributed to their higher molecular weight, which contributed to less partial volatilization/decomposition at high temperatures, especially when air is blown and a large amount of steam is formed (Huang et al., 2014; Olajide et al., 2018; Olajide et al., 2020). In addition, various authors have proven that a synergistic effect may result from the combination of two or more phenolic antioxidants linked by a covalent bond in a single molecule, which is greater than individual antioxidants (Kancheva et al., 2010; Teixeira et al., 2013). Herein, the Log P value (hydrophobicity) of the resulting compounds increased due to the presence of a methylene bridge and bulky tert-butyl moiety (an excellent electron-donating group), thus leading to substantial activity of the antioxidants in oil. The electron-withdrawing effect of the carboxylic group on GA generates unstable phenoxy radicals (Torres de Pinedo et al., 2007), however, replacing this with electron-donating molecules (Scheme 1) resulted in a well-stabilized phenoxy radical. Moreover, the excellent antioxidant capacity of the new compounds, especially that of 5b, can be attributed to the strong steric synergy exhibited separately in the form of hydrogen bonding between hydroxyl groups in the pyrogallol (Farhoosh et al., 2016) and sterically hindered hydroquinone units (Huang et al., 2014) of the molecule, a phenomenal effect which allowed the less stable free radical of 5b to easily convert to a more stable form intramolecularly. Furthermore, the pyrogallol component may easily synergistically regenerate the sterically-hindered hydroquinone one by donating hydrogen to it radical forms (Guo et al., 2017).

This method is frequently used to evaluate the potency of antioxidants as radical scavengers in that it is easily reproducible, rapid and sensitive. The scavenging activities of the antioxidants studied are shown in Figure 2. All the tested compounds showed a steady increase in scavenging activities between 1.5 and 48 µM, except GA, which exhibited a rapid one. At 12 µM, the scavenging abilities of GA, BHT, TBHQ, 5a and 5b were 67, 12, 23, 19 and 32%, respectively. Similarly, EC₅₀, the effective concentration needed to reduce the initial DPPH concentration by 50% is provided in Table 2. GA exhibited the highest EC₅₀ value (19.09 µM) while BHT expectedly showed the least (54.84 µM). The antioxidant activity of the antioxidants in the DPPH system decreased as follows: GA > 5b > TBHQ > 5a > BHT.

This finding is in agreement with previous studies (Alavi Rafiee et al., 2018; Asnaashari et al., 2014; Farhoosh et al., 2016; Mansouri et al., 2020), where GA was found to be the most potent anti-radical agent compared to a subset of other phenolic antioxidants. The ability of phenolics to scavenge radicals is dependent on the number of electron-donating hydroxyl groups in the phenyl ring, which increase phenoxy radical stability. As a result, GA, with a carboxyl and three hydroxyl substituents, was more active than 5b, TBHQ, 5a and BHT, as more hydrogen can be donated from the phenolic hydroxyls to stabilize the free radicals (de Pinedo et al., 2007). Moreover, despite the electron-withdrawing effect of the carboxyl group (-COOH), it can readily dissociate in polar media to carboxylate anion (-COO-), an electron-donating moiety which supports the formation of more stable phenoxyl radicals (Mansouri et al., 2020). The pattern of interaction between phenol and the DPPH radical occurs in the form of phenolic hydrogen abstraction and the transfer of a second hydrogen and/or formation of dimers (reactive or non-reactive) from the phenoxyl radical (Guitard et al., 2016). TBHQ, with a hydroquinone structure, can release two hydrogens to DPPH. However, the presence of tert-butyl creates a steric hindrance effect on the ortho-OH substituent and thus lowers the abstraction of hydrogen by the DPPH (Weng and Huang, 2014).

This effect was apparent in the scavenging effect of 5b, but nevertheless, the additional presence of a pyrogallol unit clearly increased its overall scavenging ability, thereby surpassing that of TBHQ. Strongly sterically hindered monophenols like BHT can only transfer one hydrogen per molecule and are unable to form reactive dimers (Guitard et al., 2016), leading to low DPPH scavenging activity compared to GA and other antioxidants studied here. Nonetheless, 5a displayed a higher scavenging activity (EC₅₀ = 33.66 µM) compared to BHT (EC₅₀ = 54.84 µM), mainly due to the presence of pyrogallol component in its molecule.

Meanwhile, the radical scavenging activity of antioxidants may also be illustrated by the degree of polarity, which determines their solubility and availability to oxidative compounds such as DPPH (Mansouri et al., 2020). Compounds with higher log P values are more hydrophobic. Therefore, based on theoretical partition coefficient (Log P_theor) values (Table 2), GA had higher polarity than other compounds studied herein, leading to better interactions between it and DPPH. Briefly, as seen in this study, molecular configuration, ability to donate hydrogen atoms and subsequently stabilize phenoxyl radicals all influenced the antioxidant activity of the different phenolic antioxidants (Silva et al., 2000).

As shown in Figure 3A, the initial PV of the samples was 1.47 ± 0.06 mEq/Kg oil, depicting that the oil was in good condition (PV < 15 mEq O₂/Kg) (Maszewska et al., 2018). As expected, BHT and control displayed a sharp increase in PV, reaching 80 mEq/kg at 4.6 and 8.5 days of storage, respectively. At the same time, GA and 5a samples oxidized slowly at the initial stage, and then steeply increased, reaching a maximum on days 12 (71.52 ± 0.85 mEq/kg) and 18 (88.20 ± 0.26 mEq/kg), respectively. However, both TBHQ and 5b had better oxidative stability by exhibiting a gradual increase in PV throughout the storage period.

A p-AV below 10 depicts good quality oil (Maszewska et al., 2018). Results from Figure 3B showed that after 30 days of storage at 65 ± 0.5 °C, most of the samples had a p-AV exceeding 10 from an initial value of 1.35 ± 0.03, except for the TBHQ (3.14 ± 0.14) and 5b (4.71 ± 0.02) groups. The control and BHT samples showed similar result trends as earlier presented in PV with p-AV of 10.14 ± 0.07 and 9.94 ± 0.07 on days 6 and 9, respectively. Nevertheless, GA extended the oxidative stability of oil samples in terms of p-AV by an additional 12 days compared to what was observed for PV, thereby reaching a maximum value of 11.22 ± 0.18 on the 24^th day of storage.

Totox considers both peroxides and aldehydes generated during oxidation and is thus a better indicator of overall oxidative deterioration (i.e. lower Totox value depicts better oil quality and vice versa). The results presented in Figure 3C are similar in trend to those of PV and p-AV under the same storage conditions except for GA, which displayed extended oxidative stability by delaying the production of secondary oxidation products for 12 more days. However, this effect showed little significance in terms of total oxidation. Nonetheless, the antioxidant capacity of the compounds evaluated in bulk oil under the Schaal Oven test decreased as follows: TBHQ ≈ 5b > 5a > GA > BHT > control.

The dissimilarity in the antioxidant activities of compounds may depend on their chemical structures, which influence their ability to stabilize their own phenoxy radicals. Compounds 5a and 5b both exhibited stronger oxidative stability than TBHQ at high temperatures (> 100 °C) under Rancimat, but at moderate thermal temperatures (65 °C) the protective activity of 5a was lower and that of 5b was close to TBHQ. This observation is in accordance with previous reports (Olajide et al., 2020; Zhang et al., 2004). The phenolic hydroxyl groups on TBHQ have strong steric synergy and can transfer hydrogen atoms to active peroxyl radicals in lipid matrices to interrupt the oxidative process. Similar to a previous study (Farhoosh et al., 2016), despite the lower polarity and DPPH scavenging activities of 5a and 5b (Figure 2; Table 2), they were more effective antioxidants than GA in bulk oil. This is probably due to their fundamental carry-through properties-i.e., the ability to resist decomposition by heat and/or loss via volatilization. Moreover, in the lipid media, hydrogen atoms from sterically hindered moieties linked to 5a and 5b may interact intramolecularly with the meta-hydroxyl substituents of their pyrogallol component, subsequently leading to the stabilization of phenoxy radicals and thus increased antioxidant activities (Asnaashari et al., 2014).

The activity of antioxidants evaluated in emulsion following accelerated oxidation in the oven until PV equaled 70 mEq/kg contrasted that observed for bulk oil, and the relative decreasing order was: 5b > 5a > BHT > TBHQ > GA > control (Figure 4). Similarly, this finding was different from the trend observed in the DPPH assay which suggests antioxidant performance depends greatly on the method of analysis and/or system used (Farhoosh et al., 2016). The effectiveness of antioxidants in the emulsion system were generally reduced compared to those in the soybean bulk oil, indicating lower antioxidant concentration in the hydrophobic center of oil droplets, thus leading to a faster oxidation rate. Gallic acid, with the highest scavenging activity under DPPH assay, was the weakest in the emulsion system. This conforms with previous studies, where the antioxidant activities of GA (Asnaashari et al., 2014) and rosmarinate alkyl esters (Panya et al., 2010; Schwarz et al., 2000) in emulsion increased as their polarity decreased. These lower polar compounds up to their dodecyl derivatives were more effective antioxidants in that they predominantly concentrated themselves at the oil-water interface where oxidation occurs. However, GA and rosmarinic acid with the highest polarity incorporated a larger amount of their molecules to the aqueous phase than the interface.

Comparably, TBHQ, with the strongest antioxidant activity in bulk oil (Figure 3), exhibited weak antioxidant activity compared to the more lipophilic BHT, 5a and 5b. Antioxidants partition on the basis of polarity, which is determined by their molecular structural properties (i.e. hydrophobic compounds display higher log P values and vice versa) (Asnaashari et al., 2014). Zhang et al., (2004) also reported weak antioxidant activity for TBHQ, which was due to its low hydrophobicity leading to more migration towards the aqueous phase in the emulsion than the oil-water interface (oxidation site).

According to Panya et al., (2010), butyl rosmarinate, an alkyl ester of rosmarinic acid, with similar molecular structure and Log P_theor value to 5b, predominantly concentrated at the interface and thus exerted stronger antioxidant capacity than the more polar rosmarinic acid and least polar eicosyl rosmarinate (20 carbon atoms). As a result, the combined hydrophobic nature of the methylene bridge and the tert-butyl substituent in 5b must have shifted the molecule to the oil-water interface, leading to its greater oxidative stability compared to other antioxidants studied herein. Meanwhile, BHT, with the highest hydrophobicity among all the phenolics studied, may have partitioned more into the oil phase away from the interface, thus exhibiting reduced effectiveness compared to 5a and 5b. This was in agreement with the results obtained by Li et al., (2006) for BHT in soybean oil-in-water emulsions. Hence, contrary to the performance in the DPPH assay, the lipophilic derivatives-5a and 5b, mostly exhibited stronger antioxidant activities than their individual parent molecules in emulsion and bulk oil systems, including the high temperature (100 - 140 °C) Rancimat experiment.

This can be attributed to the fact that alkylation of the pyrogallol unit with sterically hindered hydroquinone or 2,6-DTBP moieties in 5b and 5a molecules, respectively, decreased the electron-withdrawing effect of -COOH linked to the phenyl ring of GA, resulting in a phenoxy radical with better stabilization (Farhoosh and Nyström, 2018). Indeed, the effectiveness of phenolic antioxidants in oil indicates that their stabilizing capacity is considerably related to the properties of functional groups present, steric synergy and intramolecular hydrogen bonding. Meanwhile, in the emulsion system, it may depend on variables such as polarity and solubility, emulsifier used, concentration and type of antioxidants, radical-scavenging properties and the complex effects at the oil-water interfaces. Moreover, the polar paradox theory, which denotes that more polar antioxidants exhibit lower capacity in more polar media (Alavi Rafiee et al., 2018), is not applicable in many cases.