Kinetic Analysis of High-Temperature Sunflower Oil Peroxidation Inhibited by the Major Families of Phenolic Antioxidants Unveils the Extraordinary Activity of 1,4-Hydroquinones

Peroxidation of vegetable oils represents a major problem for the food and biodiesel industries, and it is greatly accelerated by oil degree of unsaturation and by temperature increase. Phenols represent the most common additives used to counteract oil peroxidation, however clear structure-activity relationships at high temperatures are not available. We report, herein, a kinetic study of O2 consumption during spontaneous peroxidation of sunflower oil at 130 °C in the presence of 18 antioxidants belonging to the main families of natural and synthetic phenols, including α-tocopherol, alkylphenols (BHT, BHA), hydroquinones (TBHD), catechols (quercetin, catechin) and gallates. Results show that TBHQ provide the best protection in terms of induction period (IP) duration and O2 consumption rate. EPR spectroscopy demonstrated that the inhibition activity is negatively correlated to the stability of the phenoxyl radical of the antioxidant (A•), suggesting that chain propagation with linoleate (RH) moieties A• + RH → AH + R• decreases the efficacy of those antioxidants forming persistent A• radicals. These results provide important information to optimize the antioxidant activity of phenols and of novel phenol-based materials.


Introduction
Lipid peroxidation is a main problem for edible oil utilization, and for the long-term preservation of foods rich in polyunsaturated lipids. The formation of hydroperoxides and other oxygenated products like epoxides, aldehydes, ketones, diols, etc. are correlated with off-flavor, insoluble deposits, and with the formation of potentially toxic compounds [1]. Besides, recycling used oil as feedstock for biorefineries requires the control of oxidation and polymerization processes that can cause corrosion and clogging problems [2]. It is therefore understandable why the development of novel and more effective ways to counteract lipid peroxidation is still an active research field. Phenols represent the most important class of natural and artificial antioxidants, and therefore it is not surprising that many recent studies are based on the manipulation and optimization of their characteristics. For example, in the emerging field of nano-antioxidants [3,4], natural phenolic polymers like lignin [5] and melanin-like materials [6,7] are among the most promising radical-trapping platforms. Besides, many natural-based waste products currently investigated as a source of antioxidants are particularly rich in phenols, like for instance, cashew-nut shell liquid [8], or olive mill wastewater [9].

Autoxidation Experiments
Oxygen consumption during the sunflower oil autoxidation was measured by an O 2 -sensitive optical probe (Pyroscience GmbH, Aachen, Germany) in a sealed apparatus previously described [16]. Briefly, 2.00 g of oil containing variable amounts of the antioxidants is introduced into a 10 mL round-bottomed flask with a magnetic stir bar connected to a miniaturized water-cooled condenser to keep the temperature of the O 2 -sensitive probe below 40 • C (total internal volume 29 mL). The sample is kept under vigorous stirring in a silicone bath at 130 • C. The O 2 probe is calibrated in air before each experiment. The amount of consumed O 2 is calculated from the internal volume of the apparatus. The experiments with peroxidized SSO were performed by oxidizing SSO at 130 • C measuring O 2 uptake. Then the sample was aerated, and a small volume of a concentrated octanol solution of 10 was added.

Analysis of Antioxidant Stability
The experiments were performed under the same conditions as the autoxidation experiments, except for the use of triacetin, which is resistant to oxidation at 130 • C, in place of SSO. The concentrations of the antioxidants were determined by HPLC separation coupled to a photodiode array (DAD) UV-vis detection and the flow was set to 0.9 mL/min. The column was a Sepachrom C 18

EPR Analysis
The determination of the concentration of the radicals of 2, 3, 7, and 10 was performed by an X-band electron paramagnetic resonance (EPR) spectrometer (Elexsys 500, Bruker, Billerica, MA, USA) equipped with a variable temperature dewar placed in the resonant cavity. Suprasil quartz EPR tubes were filled with the pure oil samples (no dilution) up to 4 cm and thermostatted at 40 • C by nitrogen gas flowing in the dewar. The small quantity of oil ensured a homogeneous temperature inside the analyzed sample. Typical instrumental settings were: microwave power = 10 mW, amplitude = 2 G, 64 accumulations. The radical concentration was obtained from the double integration of the EPR signal multiplied by the inverse temperature in Kelvin degrees to correct for the Curie temperature dependence and calibrated with the signal of the strong-pitch standard from Bruker. The spectra were simulated with the software Winsim [19], freely available from the Public Electron paramagnetic resonance Software Tools (PEST) from NIEHS (National Institue of Environmental Health Sciences). The radicals were identified as the phenoxyl radicals of the investigated antioxidants from the agreement of their spectral parameters (hyperfine coupling constants, a, and g-factors) with those reported in the literature [20,21] or with those expected from their structure [22].

Statistical Analysis
O 2 -consumption experiments and HPLC analysis were performed in duplicate. The values are expressed as the average ± SD (standard deviation). Student's t-test was applied for comparisons between two mean values. Differences were considered to be of significance when p was ≤ 0.05.

Autoxidation Experiments
Stripped high-linoleic sunflower oil (SSO), in the absence of antioxidants, showed an initial linear O 2 consumption, followed by a decrease in the oxidation rate with the decline of O 2 concentration in the headspace ( Figure 1, trace a). In the presence of the antioxidants, the autoxidation onset of SSO was retarded for a variable period of time (induction period, IP), after which the O2 consumption was faster. The minimum slope of O2 consumption during the inhibited or retarded period (inhibited rate, Rinh) and the duration of the inhibition IP were taken as indicators of the efficacy of the antioxidants. In Figure 1, some typical O2 consumption curves and the derivation of Rinh and IP are reported. By following this procedure, we analyzed several natural and synthetic phenolic antioxidants all at the identical concentration of 0.1% w/w, and the results are reported in Scheme 1. The antioxidants were divided into three families depending on the number of OH groups and their relative position because, as will be evident in the discussion of the results, this feature influences the fate of the radical of the antioxidant and the overall antioxidant efficacy.
Monophenols, characterized by the presence of only one phenolic OH group, include -tocopherol (7) and its synthetic analog 8, two ortho-methoxy phenols (5 and 6) having a structural relationship with natural compounds like sinapic acid and eugenol, the wellknown antioxidant additives BHT (3) and BHA (2), and other alkylated or methoxylated phenols used for comparison purposes (1 and 4).
Hydroquinones are characterized by the presence of two OH groups in para position, and a variable number of alkyl substituents. They include unsubstituted hydroquinone (9), tert-butyl hydroquinone (TBHQ, 10), and 2,5-di-tert-butyl hydroquinone (11). The corresponding oxidized para-quinones (12)(13)(14) were also investigated to ascertain the influence of these compounds on the radical chain, as it was reported that they may act as alkyl (R • ) [23] or HOO • radical traps [24].
The last group is represented by catechols, having at least two OH groups in ortho position. They include the natural antioxidants caffeic acid (18), catechin (19), and quercetin (20) as well as unsubstituted catechol (15) and the alkylated catechols 16 and 17, while gallates are represented by the antioxidant additive propyl gallate (21).
With the exception of 4-methoxyphenol (1) and of benzoquinones 12-14, all the investigated compounds had some retarding effect on SSO peroxidation. However, the only antioxidants that were able to afford a strong inhibition for a long time were the hydroquinones 9 and 10. In fact, they were able to reduce 30 and 100 times the slope of O2 consumption, respectively, while the IP was about 2.5 × 10 4 s, that is, considerably longer than In the presence of the antioxidants, the autoxidation onset of SSO was retarded for a variable period of time (induction period, IP), after which the O 2 consumption was faster. The minimum slope of O 2 consumption during the inhibited or retarded period (inhibited rate, R inh ) and the duration of the inhibition IP were taken as indicators of the efficacy of the antioxidants. In Figure 1, some typical O 2 consumption curves and the derivation of R inh and IP are reported. By following this procedure, we analyzed several natural and synthetic phenolic antioxidants all at the identical concentration of 0.1% w/w, and the results are reported in Scheme 1. The antioxidants were divided into three families depending on the number of OH groups and their relative position because, as will be evident in the discussion of the results, this feature influences the fate of the radical of the antioxidant and the overall antioxidant efficacy.
Monophenols, characterized by the presence of only one phenolic OH group, include α-tocopherol (7) and its synthetic analog 8, two ortho-methoxy phenols (5 and 6) having a structural relationship with natural compounds like sinapic acid and eugenol, the wellknown antioxidant additives BHT (3) and BHA (2), and other alkylated or methoxylated phenols used for comparison purposes (1 and 4).
Hydroquinones are characterized by the presence of two OH groups in para position, and a variable number of alkyl substituents. They include unsubstituted hydroquinone (9), tert-butyl hydroquinone (TBHQ, 10), and 2,5-di-tert-butyl hydroquinone (11). The corresponding oxidized para-quinones (12)(13)(14) were also investigated to ascertain the influence of these compounds on the radical chain, as it was reported that they may act as alkyl (R • ) [23] or HOO • radical traps [24].
The last group is represented by catechols, having at least two OH groups in ortho position. They include the natural antioxidants caffeic acid (18), catechin (19), and quercetin (20) as well as unsubstituted catechol (15) and the alkylated catechols 16 and 17, while gallates are represented by the antioxidant additive propyl gallate (21).
With the exception of 4-methoxyphenol (1) and of benzoquinones 12-14, all the investigated compounds had some retarding effect on SSO peroxidation. However, the only antioxidants that were able to afford a strong inhibition for a long time were the hydroquinones 9 and 10. In fact, they were able to reduce 30 and 100 times the slope of O 2 consumption, respectively, while the IP was about 2.5 × 10 4 s, that is, considerably longer than that of the best non-hydroquinone antioxidants like propyl gallate. If considered on a molar basis, 10 displayed the highest radical trapping stoichiometry, that is the number of radicals trapped by each antioxidant molecule, among all investigated phenols. Because of its exceptional characteristics, the IP dependence on the concentration of 10 was investigated in detail, as shown in Figure 2. The inset in this figure shows that in the concentration range considered, IP is linearly dependent on 10 concentrations. 022, 11, x FOR PEER REVIEW 5 of 14 that of the best non-hydroquinone antioxidants like propyl gallate. If considered on a molar basis, 10 displayed the highest radical trapping stoichiometry, that is the number of radicals trapped by each antioxidant molecule, among all investigated phenols. Because of its exceptional characteristics, the IP dependence on the concentration of 10 was investigated in detail, as shown in Figure 2. The inset in this figure shows that in the concentration range considered, IP is linearly dependent on 10 concentrations.

Scheme 1.
Antioxidant effect of the mono-phenols: oxygen consumption rate in the inhibited period (Rinh) and duration of the inhibited period IP. The slope without antioxidants was 70 ± 8 nmol/s. Undefined IP indicates that no induction period is present. Scheme 1. Antioxidant effect of the mono-phenols: oxygen consumption rate in the inhibited period (R inh ) and duration of the inhibited period IP. The slope without antioxidants was 70 ± 8 nmol/s. Undefined IP indicates that no induction period is present.  In order to test the effect of hydroperoxides on the IP of 10, we studied the effect of 10 (0.02%) in an SSO sample that was previously oxidized without any antioxidant, until reaching a [ROOH] ≈ 200 mM, assuming that all consumed O2 is transformed into ROOH. The results show that the antioxidant effect of 10 almost disappears (compare plot f with b, Figure 2), indicating a major effect of peroxide accumulation on IP.

EPR Experiments
To investigate the relationship between the stability of the radicals and the antioxidant activities of the investigated phenols, we measured the electron paramagnetic resonance (EPR) spectra of 2, 3, 4, and 10 at 0.1% in SSO under air. Aerated oil samples containing the antioxidants were introduced in the EPR cavity and heated at 40 °C. This temperature was chosen as the best compromise to optimize the signal-to-noise ratio. In the case of -tocopherol (7) and BHT (3), the signal arising from their phenoxyl radicals could be clearly detected (see Figure 3a,b), while in the case of 2 the signal was weak, and in the case of 10 only traces of the radical could be evidenced. The identity of the radicals was ascertained by analyzing the experimental spectra by numerical simulation as shown by the red lines in Figure 3. This procedure afforded coupling constants (a) and g factors that were in good agreement with those reported in the literature for the respective phenoxyl radicals [19][20][21][22]. The radical concentrations (see Figure 3) were in the nanomolar range and, considering the experimental error, the radical stability decreased in the order: 3 ≈ 7 > 2 >> 10. In order to test the effect of hydroperoxides on the IP of 10, we studied the effect of 10 (0.02%) in an SSO sample that was previously oxidized without any antioxidant, until reaching a [ROOH] ≈ 200 mM, assuming that all consumed O 2 is transformed into ROOH. The results show that the antioxidant effect of 10 almost disappears (compare plot f with b, Figure 2), indicating a major effect of peroxide accumulation on IP.

EPR Experiments
To investigate the relationship between the stability of the radicals and the antioxidant activities of the investigated phenols, we measured the electron paramagnetic resonance (EPR) spectra of 2, 3, 4, and 10 at 0.1% in SSO under air. Aerated oil samples containing the antioxidants were introduced in the EPR cavity and heated at 40 • C. This temperature was chosen as the best compromise to optimize the signal-to-noise ratio. In the case of α-tocopherol (7) and BHT (3), the signal arising from their phenoxyl radicals could be clearly detected (see Figure 3a,b), while in the case of 2 the signal was weak, and in the case of 10 only traces of the radical could be evidenced. The identity of the radicals was ascertained by analyzing the experimental spectra by numerical simulation as shown by the red lines in Figure 3. This procedure afforded coupling constants (a) and g factors that were in good agreement with those reported in the literature for the respective phenoxyl radicals [19][20][21][22]. The radical concentrations (see Figure 3) were in the nanomolar range and, considering the experimental error, the radical stability decreased in the order: 3 ≈ 7 > 2 >> 10.  The coupling constants (a), the g-factor, and the radical concentrations (estimated error ± 50%) are also reported.

Stability of Antioxidants
The stability at 130 °C was investigated to understand the role of phenol degradation on the duration of the inhibited period. Selected antioxidants (see Figure 4) were dissolved in triacetin, an oxidation-resistant triacylglycerol having no unsaturations, and were placed in the same reactor used in the peroxidation experiments. Then, their concentration was measured by HPLC-DAD analysis at time intervals for 6 h. The hydroquinones 9-11 were considered in order to clarify the excellent antioxidant behavior of 9 and 10 in SSO autoxidation experiments, as opposed to the relatively small efficacy of the dialkylated analog 11. The monophenol 2 and propyl gallate (21) were included for comparison purposes, while 8 was selected to investigate the behavior of -tocopherol.
We found that in all cases the concentration remained approximately constant, except for the dialkylated hydroquinone 11 and the -tocopherol analog 8, whose concentrations,

Stability of Antioxidants
The stability at 130 • C was investigated to understand the role of phenol degradation on the duration of the inhibited period. Selected antioxidants (see Figure 4) were dissolved in triacetin, an oxidation-resistant triacylglycerol having no unsaturations, and were placed in the same reactor used in the peroxidation experiments. Then, their concentration was measured by HPLC-DAD analysis at time intervals for 6 h. The hydroquinones 9-11 were considered in order to clarify the excellent antioxidant behavior of 9 and 10 in SSO autoxidation experiments, as opposed to the relatively small efficacy of the dialkylated analog 11. The monophenol 2 and propyl gallate (21) were included for comparison purposes, while 8 was selected to investigate the behavior of α-tocopherol.
We found that in all cases the concentration remained approximately constant, except for the dialkylated hydroquinone 11 and the α-tocopherol analog 8, whose concentrations, however, decreased only by 20% and 15%, respectively, after 6 h (21.6 × 10 3 s). If consid-ering that in the SSO autoxidation experiments their IP was 7.7 × 10 3 s and 6.6 × 10 3 s, respectively, even in the case of the least stable phenols 8 and 11, thermal degradation plays a negligible role in the duration of the inhibition. In addition, these experiments also demonstrated that the antioxidant loss by evaporation was negligible even in the case of the low molecular weight phenols 2 and 8. however, decreased only by 20% and 15%, respectively, after 6 h (21.6 × 10 3 s). If considering that in the SSO autoxidation experiments their IP was 7.7 × 10 3 s and 6.6 × 10 3 s, respectively, even in the case of the least stable phenols 8 and 11, thermal degradation plays a negligible role in the duration of the inhibition. In addition, these experiments also demonstrated that the antioxidant loss by evaporation was negligible even in the case of the low molecular weight phenols 2 and 8.

Discussion
The results collected herein show, in a comprehensive manner, the effect of the major families of phenolic antioxidants on the oxidation of high-linoleic sunflower oil at 130 °C. Despite their structural similarity, phenolic antioxidants display highly variable inhibition capability, as expressed by the reduction of the Rinh rate, and by the duration of the inhibited period IP. Hydroquinones 9 and 10 display the highest inhibition, while other phenols having a very similar structure, such as their methylated analogs 1 and 2, are much less active. To understand these results, reactions 1-9 occurring during SSO peroxidation have to be considered (Scheme 2) [10,25].
The initiation step (reactions 1 and 2) consists of the formation of radicals in the system with an overall rate Ri. Reactions 3-5 describe the processes occurring in not-inhibited autoxidation, where hydrogen-atom abstraction (reaction 4) occurs (mainly) at the bisallylic position of the linoleate residues and the radical chain propagates with rate constant kp [10].
The initiation process may occur either through the direct reaction of activated C-H (usually bis-allylic ones) with O2 (reaction 1) or by decomposition of hydroperoxides ROOH (reaction 2) [25,26]. Considering that only a small fraction of bis-allylic groups is consumed, the first mechanism is expected to decline only with O2 concentration, while the second mechanism is marginal at the beginning of the reaction, or in the presence of strong antioxidants, but becomes more and more important as the autoxidation proceeds and hydroperoxides are accumulated.
The Ri parameter has a great influence on IP because the faster radicals are produced in the system, the faster the antioxidants are consumed, and the IP is shortened. By applying the steady state approximation, which states that in radical chains the concentration of radical species is constant and that the overall rate of the formation of radicals must be equal to that of their disappearance, and considering Ri is constant during the inhibited period, the following relation applies: IP = n[AH] /Ri, where [AH] is the initial concentra-

Discussion
The results collected herein show, in a comprehensive manner, the effect of the major families of phenolic antioxidants on the oxidation of high-linoleic sunflower oil at 130 • C. Despite their structural similarity, phenolic antioxidants display highly variable inhibition capability, as expressed by the reduction of the R inh rate, and by the duration of the inhibited period IP. Hydroquinones 9 and 10 display the highest inhibition, while other phenols having a very similar structure, such as their methylated analogs 1 and 2, are much less active. To understand these results, reactions 1-9 occurring during SSO peroxidation have to be considered (Scheme 2) [10,25].
The initiation step (reactions 1 and 2) consists of the formation of radicals in the system with an overall rate R i . Reactions 3-5 describe the processes occurring in not-inhibited autoxidation, where hydrogen-atom abstraction (reaction 4) occurs (mainly) at the bisallylic position of the linoleate residues and the radical chain propagates with rate constant k p [10].
The initiation process may occur either through the direct reaction of activated C-H (usually bis-allylic ones) with O 2 (reaction 1) or by decomposition of hydroperoxides ROOH (reaction 2) [25,26]. Considering that only a small fraction of bis-allylic groups is consumed, the first mechanism is expected to decline only with O 2 concentration, while the second mechanism is marginal at the beginning of the reaction, or in the presence of strong antioxidants, but becomes more and more important as the autoxidation proceeds and hydroperoxides are accumulated.
The R i parameter has a great influence on IP because the faster radicals are produced in the system, the faster the antioxidants are consumed, and the IP is shortened. By applying the steady state approximation, which states that in radical chains the concentration of radical species is constant and that the overall rate of the formation of radicals must be equal to that of their disappearance, and considering R i is constant during the inhibited period, the following relation applies: IP = n[AH]/R i , where [AH] is the initial concentration of the antioxidant and n is the number of radicals trapped by each AH. At low temperatures, phenols quench ROO • by reactions 6-8 and are usually characterized by n ≈ 2 [10,25]. In the case of 10, which is one of the best antioxidants analyzed here, the IP increases linearly with 10 concentrations (Figure 2) suggesting that Ri is constant throughout the duration of the inhibited period of all three experiments shown in Figure 2, otherwise reduction in the IP should have been observed caused by the increase of hydroperoxides. On the other hand, when 10 is used as an inhibitor of an SSO sample already containing a high level of hydroperoxides, no antioxidant effect is observed, suggesting a major role of ROOH in increasing Ri.
In the case where Ri would be only due to the bulk oil's original composition (i.e., mainly the bis-allilic hydrogen concentration) in all samples, IP would be approximately the same throughout the investigated phenols. However, it turned out not to be so: IP varied significantly depending on the antioxidant (see values in Scheme 1): antioxidants causing a small reduction of O2 consumption rate (i.e., those having high Rinh) were also characterized by low IP. To explain this effect, we hypothesized that in the case of weak antioxidants also, hydroperoxides may contribute to Ri, increasing its value with respect to the pristine bulk phase. The O2 consumed during the inhibited period of weak antioxidants is converted into hydroperoxides, that in turn increase Ri, causing a shortening of IP.
Reactions 6-8 account for the action of phenolic radical-trapping antioxidants, consisting in H-atom donation to the chain-carrying peroxyl radicals ROO • , with a kinetic rate constant kinh, followed by the decay of A • with another radical (reactions 7 and 8) [10,25]. The slope of the inhibited period, Rinh, is inversely proportional to kinh. In the steady state approximation, Rin = - [10].
In addition, it must also be considered that, if the decay of the A • radical is not efficient, it can react with the substrate, propagating the oxidative chain via reaction 9. This In the case of 10, which is one of the best antioxidants analyzed here, the IP increases linearly with 10 concentrations (Figure 2) suggesting that R i is constant throughout the duration of the inhibited period of all three experiments shown in Figure 2, otherwise reduction in the IP should have been observed caused by the increase of hydroperoxides. On the other hand, when 10 is used as an inhibitor of an SSO sample already containing a high level of hydroperoxides, no antioxidant effect is observed, suggesting a major role of ROOH in increasing R i .
In the case where R i would be only due to the bulk oil's original composition (i.e., mainly the bis-allilic hydrogen concentration) in all samples, IP would be approximately the same throughout the investigated phenols. However, it turned out not to be so: IP varied significantly depending on the antioxidant (see values in Scheme 1): antioxidants causing a small reduction of O 2 consumption rate (i.e., those having high R inh ) were also characterized by low IP. To explain this effect, we hypothesized that in the case of weak antioxidants also, hydroperoxides may contribute to R i , increasing its value with respect to the pristine bulk phase. The O 2 consumed during the inhibited period of weak antioxidants is converted into hydroperoxides, that in turn increase R i , causing a shortening of IP.
Reactions 6-8 account for the action of phenolic radical-trapping antioxidants, consisting in H-atom donation to the chain-carrying peroxyl radicals ROO • , with a kinetic rate constant k inh , followed by the decay of A • with another radical (reactions 7 and 8) [10,25]. The slope of the inhibited period, R inh , is inversely proportional to k inh . In the steady state approximation, [10].
In addition, it must also be considered that, if the decay of the A • radical is not efficient, it can react with the substrate, propagating the oxidative chain via reaction 9. This reaction, although regenerating an active antioxidant molecule AH, has, however, the undesired effect of opposing reaction 6, thus contributing to the formation of hydroperoxides and increasing R inh .
Overall, this reaction scheme shows that good antioxidants must have a high k inh value (reaction 6) and must decay efficiently to avoid the onset of reaction 9.
It is well known that, in order to have high k inh values, phenols must possess electrondonating (ED) substituents in ortho and para positions with respect to the reactive OH group (see Scheme 3) [27]. ED substituents decrease the bond dissociation enthalpy (BDE) of the OH group and thus more easily render the H-atom abstraction by ROO • radicals. The substituent effect is proportional to the ED power of the substituents (OMe = OH > Me = t Bu > H) and is modulated by the bulkiness of the substituents and by the presence of intramolecular H-bonds [27]. For example, ortho-methoxy phenols have a low reactivity toward ROO • because the OH group is engaged in a strong H-bond with the OMe substituent [28]. Accordingly, phenols having only one ED substituent (like 1) or ortho-methoxy substituents (like 5 and 6), were poor inhibitors of SSO autoxidation. reaction, although regenerating an active antioxidant molecule AH, has, however, the undesired effect of opposing reaction 6, thus contributing to the formation of hydroperoxides and increasing Rinh.
Overall, this reaction scheme shows that good antioxidants must have a high kinh value (reaction 6) and must decay efficiently to avoid the onset of reaction 9.
It is well known that, in order to have high kinh values, phenols must possess electrondonating (ED) substituents in ortho and para positions with respect to the reactive OH group (see Scheme 3) [27]. ED substituents decrease the bond dissociation enthalpy (BDE) of the OH group and thus more easily render the H-atom abstraction by ROO • radicals. The substituent effect is proportional to the ED power of the substituents (OMe = OH > Me = t Bu > H) and is modulated by the bulkiness of the substituents and by the presence of intramolecular H-bonds [27]. For example, ortho-methoxy phenols have a low reactivity toward ROO • because the OH group is engaged in a strong H-bond with the OMe substituent [28]. Accordingly, phenols having only one ED substituent (like 1) or ortho-methoxy substituents (like 5 and 6), were poor inhibitors of SSO autoxidation. However, these rules seem not to apply in every case. For instance, 2 and 10 which have the same number of substituents with the same ED ability, show instead very different antioxidant effects, both in terms of Rinh and of IP. In addition, α-tocopherol (7) and its analog 8 which have the lowest BDE(OH) among the investigated phenol [27] are not the most active ones.
As stability experiments showed that the spontaneous decay of the antioxidants at 130 °C is negligible (see Figure 4), these differences must be explained by considering the interplay between reactions 6-9. EPR experiments were performed to investigate this stability. They showed that the stability of the radicals of the antioxidants increases with the number and bulkiness of the substituents in ortho and para positions to the phenolic OH in the order 3 ≈ 7 > 2, while the stability of the radical from 10 is extremely low. This result can be explained by the tendency of phenoxyl radicals to decay mainly by radical-radical addition by using the ortho and para positions, as shown in Scheme 4A [22]. If these positions contain alkyl or methoxy substituents, the decay is slow and possibly reversible.
On the other hand, the 4-hydroxylphenoxyl radical formed by 10 (see Scheme 4B) is very unstable because it decays by H-atom transfer with another radical forming the parabenzoquinone 13 and is therefore unable to propagate the oxidation via reaction 9. Interestingly, 13 is the main product that is formed in oxidized soybean oil stabilized by 10 [29].
The instability ranking experimentally observed by EPR follows the antioxidant ranking, proving that persistent radicals can behave detrimentally, because of their contribution to the chain propagation mechanism.
In this context, the disappointing behavior of catechol and gallates was unexpected, because also in their case, the phenoxyl radical may potentially decay by forming an orthoquinone. However, the present results rather suggest that the OH groups in the phenoxyl radicals from catechols and gallates are not easily transferred to another radical, likely However, these rules seem not to apply in every case. For instance, 2 and 10 which have the same number of substituents with the same ED ability, show instead very different antioxidant effects, both in terms of R inh and of IP. In addition, α-tocopherol (7) and its analog 8 which have the lowest BDE(OH) among the investigated phenol [27] are not the most active ones.
As stability experiments showed that the spontaneous decay of the antioxidants at 130 • C is negligible (see Figure 4), these differences must be explained by considering the interplay between reactions 6-9. EPR experiments were performed to investigate this stability. They showed that the stability of the radicals of the antioxidants increases with the number and bulkiness of the substituents in ortho and para positions to the phenolic OH in the order 3 ≈ 7 > 2, while the stability of the radical from 10 is extremely low. This result can be explained by the tendency of phenoxyl radicals to decay mainly by radicalradical addition by using the ortho and para positions, as shown in Scheme 4A [22]. If these positions contain alkyl or methoxy substituents, the decay is slow and possibly reversible.
On the other hand, the 4-hydroxylphenoxyl radical formed by 10 (see Scheme 4B) is very unstable because it decays by H-atom transfer with another radical forming the para-benzoquinone 13 and is therefore unable to propagate the oxidation via reaction 9. Interestingly, 13 is the main product that is formed in oxidized soybean oil stabilized by 10 [29].
The instability ranking experimentally observed by EPR follows the antioxidant ranking, proving that persistent radicals can behave detrimentally, because of their contribution to the chain propagation mechanism.
In this context, the disappointing behavior of catechol and gallates was unexpected, because also in their case, the phenoxyl radical may potentially decay by forming an orthoquinone. However, the present results rather suggest that the OH groups in the phenoxyl radicals from catechols and gallates are not easily transferred to another radical, likely because they are involved in a strong intramolecular H-bond with the -O • group (see Scheme 4C) [30].
In agreement with this view of reasoning, the plot of IP versus R inh , both divided by the molar concentration of the antioxidants to take into account their different molecular weights, shows the presence of a general trend consisting of an inverse relation between IP and R inh (see Figure 5), although the rather scattered data points suggest that other factors contribute to the IP vs. R inh relationship. In agreement with this view of reasoning, the plot of IP versus Rinh, both divided by the molar concentration of the antioxidants to take into account their different molecular weights, shows the presence of a general trend consisting of an inverse relation between IP and Rinh (see Figure 5), although the rather scattered data points suggest that other factors contribute to the IP vs. Rinh relationship. It should also be pointed out that the phenomenon of chain propagation by the phenoxyl radicals of the antioxidants A • is not observed at low temperatures when the autoxidation of SSO is accelerated by the decomposition of azoinitiators, probably because the high flux of ROO • radicals ensures an efficient decay of A • for all antioxidants, while the low temperature slows down reaction 9 [31]. On the other hand, the low IP of α-tocopherol compared to γ-tocopherol in soybean and sunflower oils at 100 °C has been in part attributed to reaction 9 [32]. Moreover, the chain propagation by the α-tocopheroxyl radicals It should also be pointed out that the phenomenon of chain propagation by the phenoxyl radicals of the antioxidants A • is not observed at low temperatures when the autoxidation of SSO is accelerated by the decomposition of azoinitiators, probably because the high flux of ROO • radicals ensures an efficient decay of A • for all antioxidants, while the low temperature slows down reaction 9 [31]. On the other hand, the low IP of α-tocopherol compared to γ-tocopherol in soybean and sunflower oils at 100 • C has been in part attributed to reaction 9 [32]. Moreover, the chain propagation by the α-tocopheroxyl radicals is responsible for the tocopherol-mediated peroxidation in low-density lipoproteins [33].
The low efficacy of the dialkylated hydroquinone 11 is instead to be ascribed to the tendency of the electron-rich semiquinones to react with O 2 forming para-benzoquinones and HOO • , which propagates the oxidative chain (Scheme 5) [34]. It should also be pointed out that the phenomenon of chain propagation by the phenoxyl radicals of the antioxidants A • is not observed at low temperatures when the autoxidation of SSO is accelerated by the decomposition of azoinitiators, probably because the high flux of ROO • radicals ensures an efficient decay of A • for all antioxidants, while the low temperature slows down reaction 9 [31]. On the other hand, the low IP of α-tocopherol compared to γ-tocopherol in soybean and sunflower oils at 100 °C has been in part attributed to reaction 9 [32]. Moreover, the chain propagation by the α-tocopheroxyl radicals is responsible for the tocopherol-mediated peroxidation in low-density lipoproteins [33].
The low efficacy of the dialkylated hydroquinone 11 is instead to be ascribed to the tendency of the electron-rich semiquinones to react with O2 forming para-benzoquinones and HOO • , which propagates the oxidative chain (Scheme 5) [34]. Scheme 5. Chain propagation of dialkylated semiquinone radicals.

Conclusions
Autoxidation experiments at 130 °C of stripped sunflower oil in the presence of various phenolic antioxidants, having monophenolic, hydroquinone, or catechol active moieties, showed that hydroquinones with one alkyl substituent display the best activity in terms of inhibition degree of O2 consumption and duration of the inhibited period. By using EPR spectroscopy, we could demonstrate that the stability of the radical of the antioxidants plays a counterintuitive role in determining the overall activity. Antioxidants like α-tocopherol, which form relatively persistent radicals, are not very effective in counteracting oil peroxidation, because their radicals can abstract an H-atom from polyunsaturated lipids. On the other hand, hydroquinones form very unstable semiquinone radicals Scheme 5. Chain propagation of dialkylated semiquinone radicals.

Conclusions
Autoxidation experiments at 130 • C of stripped sunflower oil in the presence of various phenolic antioxidants, having monophenolic, hydroquinone, or catechol active moieties, showed that hydroquinones with one alkyl substituent display the best activity in terms of inhibition degree of O 2 consumption and duration of the inhibited period. By using EPR spectroscopy, we could demonstrate that the stability of the radical of the antioxidants plays a counterintuitive role in determining the overall activity. Antioxidants like αtocopherol, which form relatively persistent radicals, are not very effective in counteracting oil peroxidation, because their radicals can abstract an H-atom from polyunsaturated lipids. On the other hand, hydroquinones form very unstable semiquinone radicals that disappear quickly by H-atom donation to another radical, and thus are incapable of propagating the oxidative chain.
Although many efforts are still needed to fully clarify the mechanisms of the inhibition of lipid peroxidation, these results are important to develop novel antioxidant strategies for the protection of oils with a high polyunsaturated fatty acid content at high temperatures, and possibly of lipid-containing foods in general. Considering the interest in finding new sources of natural phenolic antioxidants, knowledge of the structural factors underlying their activity can drive the research toward new (poly)phenolic materials having hydroquinone functionality [35] with practical applications for food and biodiesel stabilization.

Conflicts of Interest:
The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.