Effect of rosemary extract on sunflower oil degradation studied by 1H NMR: Differences under frying conditions and accelerated storage

Effect of rosemary extract on sunflower oil degradation studied by 1 H NMR: Differences under frying conditions and accelerated storage Ainhoa Ruiz-Aracama, Jon Alberdi-Cede˜ no , Barbara Nieva-Echevarria , Andrea Martinez-Yusta , Encarnacion Goicoechea-Oses * Food Technology, Faculty of Pharmacy, Lascaray Research Center, University of the Basque Country (UPV/EHU), Vitoria-Gasteiz, Spain ARTICLE INFO Keywords: Accelerated storage Antioxidant Frying conditions Primary and secondary oxidation products 1 H NMR Rosemary extract ABSTRACT The antioxidant capacity of rosemary extract (RE) has been widely studied using classical methodologies, which offer limited information. Instead, Proton Nuclear Magnetic Resonance ( 1 H NMR) informs about the degradation rate of oil components and the nature and evolution of the products formed. This study aims to investigate the effect of RE-addition (containing 0.005 % and 0.02 % of carnosol+carnosic acid) to sunflower oil on its degradation under frying conditions without food (170 ±5 ºC) and accelerated storage (70 ◦C) by 1 H NMR. In the former, changes in oil viscosity and colour were also studied. During frying, the commercial RE added did not protect the oil, being the degradation of linoleic very similar to control. In contrast, under storage, RE behaved as an antioxidant, mainly at the highest RE-concentration, delaying the degradation of oil components and the formation of oxidation products. Thus, the effect of RE-enrichment on oil degradation depends on the conditions the oil is subjected to. 1. Introduction Lipid oxidation is a very complex process that can occur during food processing and storage, leading to reduced nutritional and sensory value and the generation of potentially toxic compounds for human health (Frankel, 2005; Guill´ en and Goicoechea, 2008). One of the main strategies employed by the food industry to delay or inhibit food lipid oxidation is the addition of antioxidants. Synthetic antioxidants like butylated hydroxytoluene (BHT), butylated hydroxyanisole (BHA) and tert-butyl hydroxyquinone (TBHQ) have been predominantly used over many years. However, the growing consumer concern about their potential toxicity, coupled with increasing demand for food containing bioactive compounds, has shifted attention toward alternative antioxidants of natural origin derived primarily from vegetable sources (Wang et al., 2023). Rosemary Extract (RE) is among the most widely utilised antioxidants and, along with Tocopherol-rich extract, is one of the two vegetable extracts included in the European legislation (EU Commission, 2011). Rosemary (Rosmarinus officinalis, L.) is a woody perennial herb native to the Mediterranean region, traditionally used in culinary applications for its flavouring, antioxidant and antimicrobial properties (Senanayake, 2018). The antioxidant activity of RE is attributed to the presence of phenolic diterpenes, mainly carnosic acid and carnosol, which contain two ortho-phenolic OH groups located at the C11 and C12 positions of catechol moieties and can serve as hydrogen donors to lipid free radicals (Senanayake, 2018). The structure of carnosic acid consists of three six-membered rings (including the above-mentioned dihydric phenolic ring), and a free carboxylic acid group, whereas carnosol, a derivative of carnosic acid, contains an additional lactone ring. It must be noted that in the last years, special attention has been paid to the potential health-protective effects of cold-pressed rosemary oil, also attributed to oil phenolic content (El-Hadary et al., 2019; Ramadan et al., 2020). Several studies have reported the antioxidant activity of RE when added to oils subjected to accelerated storage conditions at 60 ◦C (Cordeiro et al., 2013; Frankel et al., 1996; Hraˇ s et al., 2000) and to frying temperatures of 180 ◦C, either in the absence (Casarotti & Jorge, 2014; Ramalho & Jorge, 2008) or in the presence of food (Che Man & Jaswir, 2000; Che Man & Tan, 1999, Li et al., 2021; Tohma & Turan, 2015; Urbanˇ ciˇ c et al., 2014). Nevertheless, it should be noted that other authors have also indicated that the antioxidant effect of RE is temperature-dependent, as high temperatures affect the stability of its phenolic components (Schwarz et al., 1992). Thus, when the effect of RE on chicken fat oxidative stability was studied at 90–150 ◦C, the best antioxidant effect was observed below the degradation temperature of RE phenolic components (130 ◦C), whereas no antioxidant effect was * Corresponding author. E-mail address: [email protected] (E. Goicoechea-Oses). Contents lists available at ScienceDirect Food Chemistry journal homepage: www.elsevier.com/locate/foodchem https://doi.org/10.1016/j.foodchem.2025.143146 Received 2 October 2024; Received in revised form 17 January 2025; Accepted 28 January 2025 Food Chemistry 474 (2025) 143146 Available online 29 January 2025 0308-8146/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/bync-nd/4.0/ ). reported above that temperature (Liu et al., 2022). Similarly, when carnosic acid was added to virgin olive oil subjected to accelerated storage at 60 ◦C, an antioxidant effect was observed. However, when exposed to frying temperatures (180 ◦C), no protective effect was evidenced, which was attributed to the decomposition of carnosic acid (Zunin et al., 2010). In most of the aforementioned studies, classical methodologies were employed to assess the effect of RE on lipid oxidation, including Peroxide Value and the measurement of Conjugated Dienes by ultraviolet absorption for primary oxidation products, p-Anisidine Value or Thiobarbituric Acid Reactive Substances (TBARS) test for secondary oxidation products; and methods to measure radical scavenging activity like DPPH (2,2 ′ -diphenylpicryl hydrazyl free radical) test. Although these assays have been widely used in recent decades due to their simplicity, it is well known that they are not comprehensive measures of lipid oxidation, because they provide limited information that, if not carefully interpreted, can lead to inconsistent conclusions (Durand et al., 2025; Frankel, 1993; Frankel & Meyer, 2000; Kerrihard et al., 2015; Stevenson et al., 1984). Thus, it is recommended to evaluate antioxidants under various oxidation conditions using specific methodologies that can identify the products formed and inhibited by the presence of antioxidants in each case (Frankel & Meyer, 2000). Over the past thirty years, Proton Nuclear Magnetic Resonance spectroscopy ( 1 H NMR) has been successfully employed to study food lipids, not only for their characterisation but also to analyse their oxidation processes under various conditions (Ampem et al., 2024; Ca˜ no-Ochoa et al., 2022a,b; Guill´ en & Ruiz, 2003, Guillen & Ruiz, 2005; Haywood et al., 1995; Hidalgo & Zamora, 2003; Hwang, 2017; Martinez-Yusta et al., 2014). 1 H NMR requires no chemical modification of the sample and can be performed in a single run of just a few minutes. It provides a substantial amount of both qualitative and quantitative information on the degradation of major and minor lipid components, as well as on the evolution of primary and secondary oxidation products, provided these are present in sufficient concentrations and their signals do not overlap with those of other components. In this context, the present study investigates, through 1 H NMR, the effect of RE addition on the degradation of sunflower oil subjected to two different processes: frying conditions in the absence of food at 170 ◦C and accelerated storage conditions at 70 ◦C. Special attention is given to the degradation of the main unsaturated acyl group (linoleic, 18:2 ω 6) and the generation and evolution of several oxidation products of different nature. In addition, for the samples exposed to frying temperatures, changes in viscosity and colour are assessed for comparative purposes with other studies. 2. Materials and methods 2.1. Samples Refined sunflower oil (S) was obtained from a local oil supplier. Its initial composition, expressed as molar percentages of main acyl groups, was determined by 1 H NMR (Guill´ en & Ruiz, 2003): linoleic 57.90 ± 0.09 %, oleic 30.46 ±0.43 % and saturated 11.54 ±0.41 %. Rosemary Extract (RE) was acquired from a food additives manufacturer. According to the label, it was prepared from a deodorised ethanolic extract of the leaves of Rosmarinus officinalis and contained 10.97 % of carnosic acid plus carnosol. Enriched oils were prepared by adding RE to the oil in proportions that resulted in final concentrations of carnosic acid plus carnosol of 0.005 % and 0.02 % (by weight). The corresponding samples were designated as S +RE0005 and S +RE002, respectively. It should be noted that the European legislation stipulates that RE is an antioxidant additive, named as E-392, which can be added to frying oils (except for olive and olive pomace oils) at a maximum concentration of 50 mg/ kg (0.005 %), expressed as the sum of carnosic acid plus carnosol (Senanayake, 2018). In other types of food products, the limit is much higher (400 mg/kg, 0.04 %), and in food colourings and flavourings, even more so (1000 mg/kg, 0.1 %) (EU Commission, 2011). These samples, enriched (S +RE0005, S +RE002) and non-enriched (S) oils, were subjected to frying conditions in the absence of food, as well as to accelerated storage conditions. Prior to heating, the Peroxide Value of enriched and non-enriched oils was determined using the portable photometer HI 83730 (Hanna Instruments, Smithfield, USA), which is an adaptation of the official method described in European Commission regulation EC2568/91. This photometer was equipped with a commercially available reagent kit (range: 0–25 meq O 2 /Kg, resolution: 0.5 meq O 2 /Kg, accuracy: ±0.5 meq O 2 /Kg). The mean value for the three fresh oil samples was 4.7 ± 0.8 meq O 2 /Kg. 2.2. Oil heating at frying conditions in the absence of food Three litres of enriched (S +RE0005 and S +RE002) and nonenriched (S) oils were heated in professional fryers (Sammic F-3, 230 V, 2.7 kW, Azkoitia, Spain) at 170 ±5 ◦C, in periods of 6 h/day, until they reached the legal usage limit, set at 25 % of Total Polar Compounds (TPC) in Spain and other countries (Firestone, 2007). The dimensions of the stainless-steel tank in the fryers were 15 cm wide x 30 cm long x 15 cm high, with an oil-air surface area of 450 cm 2 and an initial oil height of approximately 8.9 cm. Oil temperature was periodically measured with a calibrated thermometer. The tank was left uncovered and was not replenished with oil throughout the heating periods; between heating sessions the oil was maintained at room temperature in the dark, with the tank cover closed. Oil samples were taken periodically (every 6–12 h of heating) and analysed by 1 H NMR; changes in viscosity and colour were also assessed. These experiments were conducted in duplicate. 2.3. Oil accelerated storage conditions Aliquots of ten grams of each oil sample (S, S +RE0005 and S + RE002) were transferred into crystal Petri dishes of 80 mm diameter and placed in a convection oven (Memmert GmbH UN110, Schwabach, Germany), with circulating air and without stirring. The oven temperature was maintained at 70 ◦C, with a stability of 0.5 %. The Petri dishes were placed in the oven without their lids to facilitate exposure to the circulating air. The process was monitored daily by 1 H NMR until total polymerisation of the samples was reached. The experiments under accelerated storage conditions were conducted in duplicate. 2.4. Acquisition of 1 H NMR spectra and derived quantitative data The 1 H NMR spectra were acquired in duplicate using a Bruker Avance 400 spectrometer operating at 400 MHz (Bruker Scientific Instruments, Billerica, USA). For sample preparation, 175 μ L of oil, either subjected to frying conditions or accelerated storage, was mixed in a 5 mm diameter 1 H NMR tube with 425 μ l of deuterated chloroform (CDCl 3 , 99.8 %), containing a small proportion of non-deuterated chloroform and 0.03 % of tetramethylsilane (TMS). This was used as reference compound to calibrate the chemical shift at 0.0 ppm (Euroisotop, Paris, France). The acquisition parameters were as follows: spectral width 6250 Hz, relaxation delay 3 s, number of scans 64, acquisition time 2.621 s and pulse width 90◦, with a total acquisition time of 6 min and 20 s. The assignment of the 1 H NMR signals (based on their chemical shifts and multiplicities) was performed using standards and/or the literature, as previously described and indicated in supplementary Table S1. The compound Z3-nonenal was acquired from Larodan (Malm¨ o, Sweden). The relaxation delay and acquisition time used to obtain the 1 H NMR spectra ensured complete relaxation of the sample protons, allowing the signal area to be proportional to the number of protons generating them. This enabled the use of 1 H NMR signals to estimate the molar percentage of linoleic acyl groups and the concentration of several oxidation products in the samples, as described in the Supplementary Material. A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 2 However, it was not possible to monitor the evolution of any compound present in RE because, at the tested concentrations, their signals were not detectable in the spectra of the enriched oils. The 1 H NMR spectra illustrating the various figures were plotted at a fixed absolute intensity value to ensure validity for comparative purposes and processed using the MNova program (Mestrelab Research, Santiago de Compostela, Spain). 2.5. Percentage in weight of Total Polar Compounds (TPC) in oils subjected to frying temperatures As previously mentioned, frying oils can be used until they reach the legal limit of 25 % TPC. This parameter was measured using the Testo 270 instrument (Testo, Lenzkirch, Germany), which determines the dielectric value of the oil by immersion in hot oil and provides the percentage by weight of TPC along with the temperature. The accuracy values are ±2.0 % TPC and ±1.5 ◦C, respectively, and the results are considered well correlated with those obtained through chromatographic determination of TPC (Osawa et al., 2012). 2.6. Other analyses of oils subjected to frying temperatures In order to obtain additional information, other parameters were measured in oils exposed to frying temperatures. The viscosity of 500 mL oil samples was determined using a rotational Viscosimeter (Visco Star Plus L, Fungilab, Barcelona, Spain) at a constant temperature of 20 ◦C and a shear rate of 200 rpm. Colour changes in the oil samples were assessed with a chromameter (CR-400, Konica Minolta, Japan), set on the CIELab colour system. The device was calibrated with a white plate prior to use. The parameter L* measures brightness, ranging from 0 (black) to 100 (white), while a* and b* are opponent chromaticity coordinates representing red-green (a* positive: red, a* negative: green) and yellow-blue (b* positive: yellow, b* negative: blue) scales, respectively (Maskan, 2003). 2.7. Statistical analysis The degradation rates shown in Table 1 and the graphical representation of several quantitative determinations (Figs. 1, 3, 4, 6 and 7) were generated using Microsoft Office Excel 2016. The standard deviations obtained for each value are not represented in the figures but are given in the Supplementary Tables S2-S11, together with the mean values of the concentrations. These tables also provide the significance of differences of the several determinations, which were calculated by one-way analysis of variance (ANOVA) followed by Tukey’s b test at p < 0.05, using SPSS Statistics v.28 (IBM, NY, USA). As can be observed in Supplementary tables, in general, when the concentration of the oxidation compounds was higher than 0.5 mmol/mol of acyl groups (AG), the standard deviation between duplicates was lower than 5 %. On the other hand, when the concentration was lower than that value, the standard deviation between duplicates could be higher than 5 %. 3. Results and discussion 3.1. Effect of RE addition on sunflower oil degradation under frying conditions Non-enriched (S) and enriched (S +RE0005, S +RE002) sunflower oil samples were heated at 170 ◦C until all reached the legal limit of 25 % TPC after 138 h, at which point the experiment was concluded (see supplementary Fig. S1 and Table S2). The changes occurring in the oil were monitored by 1 H NMR, as well as by viscosity and colour determination, with the most significant observations outlined below. 3.1.1. Effect of RE addition on the degradation of linoleic acyl groups It is well known that oil unsaturated acyl groups (AG) degrade under thermo-oxidative conditions, being the degradation level greater the higher the unsaturation degree. This degradation is illustrated in Fig. 1a, which shows the evolution of the molar percentage of linoleic chains in the three samples, as estimated by 1 H NMR. The pattern is very similar in both the absence (S) and presence of RE in the oil (S +RE0005, S + RE002). In all samples, the initial molar percentage of linoleic was 57.90 ±0.09 %, and after 138 h of heating, it decreased to 47.42 ±0.14 % (see quantitative data in supplementary Table S3). Based on this evolution of linoleic acyl groups, it can be concluded that the RE-enrichment did not exert any protective effect to sunflower oil under these conditions. In contrast, other studies have described the evolution of Iodine Value (IV) during the frying process in rosemary-oleoresin-enriched and nonenriched palm olein, reporting a significantly lesser decrease in the enriched oil (Che Man & Jaswir, 2000; Che Man & Tan, 1999). 3.1.2. Effect of RE addition on the generation of oxidation compounds When oils and fats are subjected to high temperatures in the presence of oxygen, such as during frying, a complex series of oxidative and thermal reactions take place. Under these conditions, hydroperoxides containing conjugated double bounds (also known as primary oxidation products) are formed very rapidly, and their decomposition is even faster (Dobarganes & M´ arquez-Ruiz, 2007; Frankel, 2005; Guillen & Uriarte, 2009; Stevenson et al., 1984). This decomposition leads to the formation of stable oxidation compounds of diverse nature, characterised by the presence of one or more oxygenated functional groups (Frankel, 2005). Some of these, referred to as secondary or further oxidation products, produce signals in the 1 H NMR spectra that enable their detection and, if not overlapped with other signals, their quantification. Fig. 2 shows enlarged 1 H NMR spectral regions, where signals related to several oxidation compounds can be observed. The Fig. 1. Evolution of the molar percentage of linoleic acyl groups in non-enriched sunflower oil (S) and in sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S +RE002): a) submitted to frying conditions in the absence of food (170 ◦C), b) submitted to accelerated storage conditions (70 ◦C). A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 3 assignment of these signals is provided in Table S1. 3.1.2.1. Monohydroperoxy conjugated Z,Eand E,E-octadecadienoates at frying conditions. The first oxidation products formed in the oxidation of oils rich in linoleic acyl groups are Z,E-conjugated octadecadienoates containing a hydroperoxide group (mHPO-c(Z,E)-dE), also known as primary oxidation products. These isomerise into mHPO-c(E,E)-dE, at rates that depend on factors such as temperature and the presence of antioxidants, among others. When found in sufficient concentrations, the hydroperoxide proton signal, as well as those corresponding to the associated conjugated dienic protons, become visible in the 1 H NMR spectra. Incipient signals due to to the hydroperoxide proton (signal c) of mHPO-c(Z,E)-dE, and those from their associated conjugated dienes (signal a), were visible in the spectra of S, S +RE0005 and S +RE002 samples before heating (only shown in the enlarged spectral regions of S sample at 0 h in Fig. 2). In contrast, signals corresponding to the associated conjugated dienes of mHPO-c(E,E)-dE were not detected. This indicates that the fresh oil had undergone some autoxidation, although to a minimal extent. Indeed, as indicated in the Materials and Methods section, the Peroxide Value of the initial oils was within the legal limits for fresh sunflower oil (4.7 ±0.8 meq O 2 /Kg). Once the samples were heated to frying temperatures, the signals corresponding to mHPO-c(Z,E)-dE disappeared, becoming absent or almost absent in the spectra of all samples throughout the process. This is consistent with the above-mentioned instability of these primary oxidation compounds. Other authors have examined the evolution of the Peroxide Value of palm olein enriched with rosemary-oleoresin at 0.02 % (Che Man & Tan, 1999), 0.4 % (Che Man & Jaswir, 2000), and with RE at various concentrations (Guo et al., 2016), as well as soybean oil enriched with RE at 0.02 % (Li et al., 2021) during potato chips deepfrying at 180 ◦C. However, all of them reported a protective effect on oil stability. Notably, in all these studies, Peroxide Values above 10 meq O 2 /kg were reported. Therefore, if mHPO-c-dE were present in similar concentrations in the samples studied here, they would have been detected by 1 H NMR, because in fresh oils (0 h) lower peroxide values were measured and their signals were visible in the spectra. 3.1.2.2. Secondary or further oxidation products under frying conditions. The decomposition of primary oxidation compounds leads to the formation of numerous secondary or further oxidation products of varying nature. The evolution of some of these products during the heating Fig. 2. Some 1 H NMR spectral regions of non-enriched sunflower oil (S) and sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S +RE002) after being submitted to frying conditions in the absence of food (170 ◦C) for 0, 36, 90 and 138 h. Regions have been properly enlarged for comparative purposes. The letters of the signals correspond to those indicated in Table S1. a: monohydroperoxy conjugated Z,E-octadecadienoates; c: hydroperoxyde group; d: monoketo conjugated E,E-octadecadienoates; f: secondary alcohols; g: primary alcohols; i: E2-alkenals; j: E,E-2,4-alkadienals; k: 4,5-epoxy-E2-alkenals; l: 4-hydroxyE2-alkenals; m: 4-hydroperoxy-E2-alkenals; n: E,Z-2,4-alkadienals; o: Z3-alkenals; p: alkanals; r: 4-oxo-alkanals. A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 4 process, particularly those whose concentration is influenced by oil enrichment with RE, is discussed below. a. Monoketo conjugated E,E-octadecadienoates Monoketo conjugated octadecadienoates are well-known oxidation products that can exhibit Z,Eor E,Eisomerism. These compounds were detected in very low concentrations in all sunflower oil samples subjected to frying conditions in the absence of food. However, due to the low intensity of signals from Z,E-isomers, only E,E-isomers were quantified, with their evolution shown in Fig. 3 (see quantitative data in supplementary Table S4). As can be observed, the presence of RE resulted in significantly lower concentrations of mKO-c(E,E)-dE, with concentrations in the S +RE002 sample slightly lower than in S + RE0005. This finding might have health implications, as these compounds have been reported to be cytotoxic (Zhao et al., 2015). b. Aldehydes Among the various products resulting from hydroperoxide decomposition, aldehydes are particularly relevant, with some having a strong impact on the sensory properties of oil. In the spectra of S, S +RE0005 and S +RE002 samples before heating (0 h), small signals corresponding to the aldehydic protons of alkanals (signal p) and E2-alkenals (signal i), as well as incipient signals of 4-hydroperoxy-E2-alkenals (signal m) and E,E-2,4-alkadienals (signal j), are already visible. These are only shown in the enlarged spectral region of the S sample at 0 h in Fig. 2. Furthermore, a very small triplet at 9.64 ppm is observed (signal o), which can be attributed to Z3-alkenals, in agreement with the spectrum of the standard compound Z3-nonenal and with 1 H NMR spectral data reported in the literature (Leikauf et al., 1995). To our knowledge, this is the first report of this compound being detected in bulk oils by 1 H NMR, although it is well described in plants, where it derives from enzymatically generated hydroperoxides and serves a precursor to other oxidation products, including 4-hydroperoxy-E2-nonenal and 4-oxo-E2nonenal (Gardner & Grove, 1998). Throughout heating at frying temperatures, the types and relative proportions of the aldehydes generated in the S sample are consistent with those previously detected by 1 H NMR in sunflower oil heated under similar conditions (Guillen & Uriarte, 2009; Haywood et al., 1995). These aldehydes primarily include E,E-2,4-alkadienals, E2-alkenals and alkanals, along with lower concentrations of E,Z-2,4-alkadienals, 4-hydroxy-E2-alkenals, 4-oxoalkanals and 4,5-epoxy-E2-alkenals. Fig. 4 illustrates the evolution of the concentration of these aldehyde types in S, S +RE0005 and S +RE002 oils over 138 h (see quantitative data in supplementary Tables S5 and S6). Similar aldehydes were generated in all samples, regardless of RE enrichment. For most aldehydes, similar formation rates and final concentrations were detected among the samples throughout the degradation process. However, the addition of RE provoked a significantly lower formation of E,E-2,4-alkadienals, with maximum values of approximately 1.8 mmol/mol AG in the enriched samples, compared to 2.3 mmol/mol AG in the non-enriched oil. A similar trend was observed regarding the effect of RE on the evolution of E,Z-2,4-alkadienals, which could be expected because these are precursors of the E,E-isomers (Frankel, 2005). Although this effect was moderate, it aligns with previous studies reporting lower p-Anisidine Values in RE-enriched frying oils (Che Man & Jaswir, 2000; Guo et al., 2016; Li et al., 2021; Tohma & Turan, 2015), which employs a reagent that reacts mainly with unsaturated aldehydes like 2,4-alkadienals and 2-alkenals. c. Other degradation products Another kind of secondary or further oxidation products formed during frying are alcohols, which can be either primary or secondary (Frankel, 2005; Hwang et al., 2020; Xia & Budge, 2017). These compounds are generated from the onset of heating at 170 ◦C in all samples. Due to significant signal overlap, quantification is not feasible. Nevertheless, analysis of the 1 H NMR spectra indicates that under these conditions secondary alcohols are formed in higher concentration than the primary ones, the generation of the former appearing to be enhanced in the presence of RE (see Fig. 2, signal f). In addition, some other small signals became visible in the spectra along the process, but no effect of RE-enrichment was observed on their evolution (data not shown). These signals included those related to monohydroxy conjugated Z,E-octadecadienoates (Martinez-Yusta et al., 2014), and several unidentified signals at 3.06, 3.80, 4.74 and 4.81 ppm, as well as in the region between 5.6 and 6.4 ppm. 3.1.3. Effect of RE addition on oil viscosity and colour Numerous studies in the literature evaluate the antioxidant capacity of extracts or compounds added to oils for frying using physicochemical measurements (Che Man & Jaswir, 2000; Che Man and Tan, 1999; Li et al., 2021; Urbanˇ ciˇ c et al., 2014). In the present work, the evolution of viscosity and colour during heating episodes was also assessed. As for changes in oil viscosity at frying temperatures, both enriched and nonenriched oils exhibited an increase over time, as expected (see supplementary Fig. S2 and supplementary Table S7). This increase is attributed to oil polymerisation and the formation of high-molecular weight compounds, serving as a reliable indicator of oil deterioration (Maskan, 2003; Stevenson et al., 1984). Concerning the effect of RE on oil viscosity during heating, a similar evolution was observed irrespective of the presence of RE. However, some authors have reported a lower increase in viscosity in RE-enriched frying oils compared to non-enriched ones (Che Man & Jaswir, 2000; Che Man & Tan, 1999; Li et al., 2021; Tohma & Turan, 2015). Concerning the changes occurring in oil colour under frying temperatures, supplementary Fig. S3 shows the evolution of CIELab coordinate values for enriched and non-enriched oils. During heating, the non-enriched sunflower oil (S) exhibited a decrease in the L* parameter, indicating reduced luminosity, whereas a* and b* parameters increased, consistent with previous studies (Che Man & Tan, 1999; Maskan, 2003). The increase in a* from negative to positive values implies that after 138 h of heating, the oil became redder and less green, likely due to the degradation of chlorophylls (Maskan, 2003). Furthermore, the increase in b* values during heating (always positive) denoted a more intense yellow colour, which is related to the observed loss of luminosity. As can be observed in Fig. S3, the addition of RE to sunflower oil mitigated these colour changes during heating at 170 ◦C, particularly at the highest concentration (S +RE002), aligning with previous studies (Guo et al., 2016; Tohma & Turan, 2015; Urbanˇ ciˇ c et al., 2014). However, several authors have noted that evaluating frying oil quality solely based on colour changes is not reliable (Che Man & Tan, 1999; Maskan, 2003; Urbanˇ ciˇ c et al., 2014). In summary, the addition of RE to sunflower oil had minimal impact on the degradation process of the oil under frying conditions, with all samples (enriched and non-enriched) reaching 25 % TCP at equal Fig. 3. Evolution of the concentration of monoketo conjugated E,E-octadecadienoates (mKO-c(E,E)-dE) in non-enriched sunflower oil (S) and in sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S + RE002) submitted to frying conditions at 170 ◦C. A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 5 heating times. This lack of effect may be attributed to the degradation of RE phenolic components at temperatures above 120–130 ◦C, as indicated by some authors (Liu et al., 2022; Schwarz et al., 1992); however, other authors reported an antioxidant effect of RE under frying conditions. It must be noted that most of those results are not comparable to the ones hereby presented, because in those studies the total amount of carnosic acid and carnosol present in the extracts was not considered (Casarotti & Jorge, 2014; Che Man & Jaswir, 2000; Che Man & Tan, 1999; Li et al., 2021; Ramalho & Jorge, 2008; Tohma & Turan, 2015). Among them only Urbancic et al. (2014) indicated the amount of these phenolics in RE added to sunflower oil and used for potato deep frying (0.0049 %), but in that work the authors studied oil thermooxidation using only non-specific methodologies (TPC, conjugated dienes and trienes and colour). In the study hereby presented, only slight differences were found between the enriched and non-enriched samples, with lower concentrations of E,E-2,4-alkadienals and monoketo conjugated E, E-octadecadienoates generated in the enriched oils. Moreover, the addition of RE ameliorated colour changes in oil, particularly at the highest enrichment level. In contrast, RE-enrichment slightly favoured the formation of secondary alcohols. 3.2. Effect of RE addition on sunflower oil degradation under accelerated storage It is known that the impact of adding compounds with potential antioxidant activity on the degradation process of edible oils depends on various factors, including the conditions to which the system is subjected. Since the addition of RE did not exert a clear protective effect on sunflower oil at frying temperatures (170 ◦C), non-enriched and enriched samples (S, S +RE0005 and S +RE002) were subjected to accelerated storage conditions at 70 ◦C with aeration to evaluate the antioxidant effect under mild degradative conditions. The degradation process at 70 ◦C was monitored by 1 H NMR until total polymerisation, focusing on the degradation of linoleic acyl groups and the formation and progression of oxidation products. 3.2.1. Effect of RE addition on the degradation of linoleic acyl groups Similar to the degradation observed during frying, the unsaturated acyl groups in sunflower oil also degraded during accelerated storage. Fig. 1b illustrates the molar percentage evolution of linoleic acyl groups over time in the three samples (S, S +RE0005 and S +RE002) under accelerated storage conditions (see quantitative data in supplementary Table S8). However, under these conditions, linoleic chains degradation followed four linear stages, with the length and slope of each sample detailed in Table 1. In all three samples, the slope of the first stage was minimal, evidencing an initial induction period (lag phase) with negligible degradation of linoleic acyl groups. The length of this stage in the S sample (day 0–2) is similar to that of S +RE0005 (day 0–3), but significantly shorter than in S +RE002 (day 0–8). This suggests that the addition of RE, particularly at the highest concentration, significantly delayed linoleic groups degradation under these conditions. Following this period, a second stage occurred. Its duration was similar in all three samples, although the degradation rate in the RE-enriched oils remained slightly lower than in the control. A third stage followed, characterised by a significant increase in the degradation rate of linoleic groups. Unlike the previous stages, the degradation rate in the control was lower (1.04 % per day) than in the RE-enriched samples, with no differences between the enriched samples (around 1.30 % per day in both). This stage ended when the remaining concentration of linoleic groups in all samples reached approximately 14.5 % (see Fig. 1b). From this point onwards, a fourth stage with a lower degradation rate of linoleic was observed. This stage lasted for two days in all samples and ended with their total polymerisation. In summary, the evolution of linoleic acyl groups concentration during the accelerated storage clearly showed that RE exerts a protective effect on sunflower oil oxidation, significantly extending the duration of the first stage (lag phase). This enrichment, particularly at the highest concentration of RE, markedly slowed the degradation rate of linoleic acyl groups compared to the control sample, allowing the linoleic groups to remain almost undegraded for a longer period (2 days in S, 3 days in S +RE0005, and 8 days in S +RE002). The protective effect of RE observed under accelerated storage conditions contrasts with the lack of Fig. 4. Evolution of the concentration of a) non‑oxygenated aldehydes and b) oxygenated aldehydes in non-enriched sunflower oil (S) and in sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S +RE002) submitted to frying conditions at 170 ◦C. A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 6 effect observed when the extract was added to sunflower oil subjected to frying temperatures. Nevertheless, it should also be noted that after the lag phase, the remaining process duration was almost the same regardless of whether the samples were enriched. Thus, while the addition of RE extended the lag phase and delayed the degradation of linoleic groups, once degradation progressed at a significant rate, the process evolved similarly, independent of the enrichment level. The extension of the lag phase induced by RE in the oxidation process of sunflower oil contrasts with the effects observed when other compounds with potential antioxidant capacity were added to this and other oils subjected to accelerated storage at 70 ◦C. For instance, hydroxytyrosol acetate also exhibited antioxidant capacity by decreasing the degradation rate of linoleic acyl groups, but it was not able to detain it as RE did (Ca˜ no-Ochoa et al., 2022b). In contrast, alpha-tocopherol did not exert a protective effect during the first stage of linoleic degradation. Instead, it reduced the lag phase, accelerating degradation and thus acting as a prooxidant (Ca˜ no-Ochoa et al., 2022a). 3.2.2. Effect of RE addition on the generation of oxidation compounds Under mild temperatures and in the presence of oxygen, which are conditions that occur during storage, a wide variety of oxidative Table 1 Degradation rates (D R ) of linoleic acyl groups, expressed in molar percentage/ day, in RE-enriched and non-enriched sunflower oils submitted to accelerated storage at 70 ◦C with aeration, in each one of their stages. The degradation rates agree with the slopes of the linear equations that relate the concentration of this acyl group and the storage time. The correlation coefficients of these lines are given in brackets. 1st stage (lag phase) 2nd stage 3rd stage 4th stage Samples Days D R Days D R Days D R Days D R S 0–2−0.13 (0.99) 2–5−0.67 (0.99) 5–9−10.39 (0.99) 9–11 −3.15 (0.99) S +RE0005 0–3−0.17 (0.94) 3–7−0.45 (0.97) 8–11 −13.10 (0.99) 11–13 −2.56 (0.99) S +RE002 0–8−0.06 (0.95) 8–12 −0.43 (0.97) 13–16 −13.02 (0.99) 16–18 −2.18 (0.94) Fig. 5. Some 1 H NMR spectral regions of non-enriched sunflower oil (S) and sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S +RE002) after being submitted to accelerated storage conditions (70 ◦C) for several days. Regions have been properly enlarged for comparative purposes. The letters of the signals correspond to those indicated in Table S1. a: Monohydroperoxy conjugated Z,E-octadecadienoates; b: Monohydroperoxy conjugated E,E-octadecadienoates; c: Hydroperoxyde group; d: monoketo conjugated E,E-octadecadienoates; f: secondary alcohols; g: primary alcohols; h: 2,3-epoxy-alkanals; i: E2-alkenals; j: E,E-2,4-alkadienals; k: 4,5-epoxy-E2-alkenals; l: 4-hydroxy-E2-alkenals; m: 4-hydroperoxy-E2-alkenals n: E,Z-2,4-alkadienals; p: alkanals; q: 4-oxo-E2-alkenals; s: Z2-alkenals. A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 7 compounds are generated (Frankel, 2005). As demonstrated in numerous previous studies, many of these compounds can be detected in 1 H NMR spectra of autoxidised oils (Alberdi-Cede˜ no et al., 2020; Ca˜ noOchoa et al., 2022a,b; Goicoechea & Guillen, 2010; Guillen & Ruiz, 2005), and the formation and evolution of some of these will be discussed below 3.2.2.1. Monohydroperoxy conjugated Z,Eand E,E-octadecadienoates during accelerated storage. As can be observed in Fig. 5, the intensity of signals corresponding to the hydroperoxide proton (signal c) of both mHPO-c(Z,E)-dE and of mHPO-c(E,E)-dE, as well as those related to their conjugated dienic systems (signals a and b, respectively), increaseed significantly during the process under accelerated storage conditions. As with linoleic acyl groups, the evolution of the concentration of both isomers in the different oil samples during storage was estimated from the area of the corresponding spectral signals, with their evolution shown in Fig. 6a (see quantitative data in supplementary Table S9). It can be observed that in the non-enriched sunflower oil sample (S), the concentration of both isomers increased slowly from the beginning of the process until day 5, when the second stage of linoleic group degradation also ended. From this day onwards, their concentration increased drastically until it reached a maximum on day 7, after which it decreased sharply, in such a way that by day 9 hydroperoxides were no longer detected in the sample. In RE-enriched samples, the evolution of these primary oxidation products differed somewhat. Until day 3 in S +RE0005 and day 7 in S + RE002, the initial concentration of mHPO-c(Z,E)-dE remained almost constant, coinciding with the extension of the first phase of linoleic acyl group degradation (lag phase; see Table 1). Furthermore, the formation of the E,Eisomers was slightly delayed in comparison to the control and it also remained very low during this initial period. These results are consistent with those reported by other authors, who observed that under accelerated conditions at 60–62 ◦C, the enrichment of different edible oils with 0.02 %–0.20 % of RE significantly delayed hydroperoxide formation, as determined by the Peroxide Value (Chen et al., 2014; Cordeiro et al., 2013; Frankel et al., 1996; Hraˇ s et al., 2000, Yang et al., 2016). Once this period ended, the concentration of both isomers increased slowly up to day 7 in S +RE0005 and day 12 in S +RE002. Notably, when this slow increase concluded, the concentration of both isomers was similar in the two enriched samples (18.5 and 18.9 mmol/mol AG in S +RE0005 and S +RE002, respectively) and closely matched that of the non-enriched sample (S) on day 5 (19.8 mmol/mol AG). From this point, there is a sharp increase in their concentration in both enriched samples until the maximum concentration was reached (day 9 in S + RE0005 and day 14 in S +RE002), decreasing markedly thereafter. It should be noted that the above-mentioned significant rise in the concentration of hydroperoxides occurred in the enriched samples, as it also did in the non-enriched one, at the end of the 2nd stage of the linoleic groups degradation described above. Regarding the maximum concentration reached, differences were only observed for mHPO-c(E,E)-dE, which was significantly lower the higher the enrichment level was. The delaying effect of RE-enrichment on the formation of mHPO-cdE, along with the lower maximum concentration of the E,E-isomers reached in the RE-enriched samples, further highlights the protective effect of this extract on the oxidation process of sunflower oil under accelerated storage. 3.2.2.2. Secondary or further oxidation products during accelerated storage. As a result of the decomposition of mHPO-c-dE, numerous secondary or further oxidation products have been described in the oxidation process of oils. The evolution of some of the most representative compounds in RE-enriched and non-enriched sunflower oil samples subjected to accelerated storage conditions at 70 ◦C will be discussed below. a. Monoketo conjugated Z,Eand E,E-octadecadienoates Monoketo conjugated dienes exhibiting both Z,E-and E,E-conjugated configurations (mKO-c(Z,E)-dE and mKO-c(E,E)-dE, respectively) were generated in all samples (see Fig. 5), with their concentrations exceeding those observed under frying conditions (see Fig. 2). These compounds are well-established oxidation compounds derived from mHPO-c-dE. As shown in Fig. 6b, which depicts the evolution of the concentration of both isomers, these compounds were not detected initially but appeared once their precursors reached a certain threshold (see Fig. 6. Evolution of the concentration of certain oxidation products in non-enriched sunflower oil (S) and in sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S +RE002) submitted to accelerated storage at 70 ◦C: a) monohydroperoxy conjugated octadecadienoates with Z,E- (mHPO-c(Z,E)-dE) or E,E-isomerism (mHPO-c(E,E)-dE), and of b) monoketo conjugated octadecadienoates with Z,E- (mKO-c(Z,E)-dE) or E,E-isomerism (mKO-c(E,E)-dE). A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 8 quantitative data in Supplementary Table S9). In all samples, mKO-c(Z, E)-dE were generated slightly earlier than the E,E-isomers, with this delay being more pronounced in the enriched oils, where the E,E-isomers were only detected after the second stage of linoleic degradation had concluded (see Table 1). Furthermore, the formation of mKO-c-dE in the enriched oils occurred later than in the non-enriched sample, specially at the highest RE concentration. This trend was consistent with the patterns observed for linoleic degradation and mHPO-c-dE formation described above. b. Aldehydes The spectra of S, S +RE0005 and S +RE002 prior to exposure to oxidative conditions showed signals corresponding to alkanals (signal p), E2-alkenals (signal i), 4-hydroperoxy-E2-alkenals (signal m) and E,E2,4-alkadienals, along with the newly identified Z3-alkenals (signal o) (see Fig. 5), as previously noted in the study under frying conditions. The evolution of the concentration of these and other types of aldehydes detected in samples subjected to accelerated storage conditions at 70 ◦C is presented in Fig. 7 (see quantitative data in Supplementary Table S10 and S11). In the non-enriched oil sample (S), the concentration of those aldehydes that were initially present remained almost unchanged until day 6. From that day on, their concentration increased, and simultaneously, new aldehydes were generated, as can be observed in Figs. 5 and 7. It should be noted that, among all of them, 4-hydroxy-E2-alkenals reached the highest concentration, while their precursors, 4-hydroperoxy-E2-alkenals, also attained significant concentrations. These later were not detected at frying temperatures, as described previously. It is also remarkable that, unlike what occurs under frying conditions in the absence of food, where 4-oxoalkanals are formed, under mild-oxidative conditions (as those employed in the accelerated storage study), only 4oxo-E2-alkenals were detected. A possible explanation for this observation is that, when the sample is exposed to prolonged periods at high temperatures, 4-oxo-E2-alkenals are directly reduced to the saturated 4oxo-alkanals, as postulated by Wann et al. (2021). Under these conditions, the 4-oxo-E2-alkenals formed during accelerated storage remain stable and hence, accumulate. In contrast, at frying temperatures, they are rapidly modified into 4-oxo-alkanals, and are therefore not detected. Furthermore, the generation of 2,3-epoxy-alkanals and of Z2-alkenals under accelerated storage conditions was also notable, as are the lower proportions of both E,Eand E,Zisomers of 2,4-alkadienals formed compared to those observed under frying temperatures. When sunflower oil is enriched with RE, the generation rate of aldehydes is affected in a concentration-dependent manner, as previously described for the generation of other oxidation products. Hence, in sample S +RE0005, changes in the concentration of aldehydes are observed on day 9 (3 days later than in S), whereas in sample S +RE002, this occurs on day 14 (8 days later than in S), coinciding in both cases with the time when the samples reach the maximum concentration of mHPO-c-dE. From these days onwards, the evolution of the concentration of all aldehydes closely resembles that observed in the non-enriched samples (see Fig. 7), with no differences detected between enriched and non-enriched oils in terms of the maximum concentrations of aldehydes reached. Still, the fact that the enrichment of sunflower oil with RE delays the formation of aldehydes is of great importance, since oxygenatedα ,β-unsaturated aldehydes (such as 4-hydroperoxy-E2alkenals, 4-hydroxy-E2-alkenals, 4,5-epoxy-E2-alkenals and 4-oxo-E2alkenals) are highly reactive nucleophiles that have been widely described as potential causative agents of several degenerative diseases, including Alzheimer’s, Parkinson’s, cancer and diabetes (Guill´ en & Goicoechea, 2008; Haywood et al., 1995). These results align with those obtained by other authors, who reported lower p-Anisidine Values and Thiobarbituric Acid Reactive Substances (TBARS) values in RE-enriched oils subjected to accelerated storage at 60 ◦C (Chen et al., 2014; Hraˇ s et al., 2000). c. Other degradation products In recent years, using 1 H NMR, many oxidation products of different nature have been described to be generated in oils subjected to the same accelerated storage conditions as those employed in the present study (Alberdi-Cede˜ no et al., 2020; Ca˜ no-Ochoa et al., 2022a,b). Among them, primary and secondary alcohols were found to be formed in the samples subject of this study. However, unlike what is observed at frying Fig. 7. Evolution of the concentration of a) non‑oxygenated aldehydes and b) oxygenated aldehydes in non-enriched sunflower oil (S) and in sunflower oil enriched with rosemary extract at 0.005 % (S +RE0005) and at 0.02 % (S +RE002) submitted to accelerated storage at 70 ◦C. A. Ruiz-Aracama et al. Food Chemistry 474 (2025) 143146 9