Enzymatic interesteri ﬁ cation to produce zero-trans and dialkylketones-free fats from rapeseed oil

– This study aims to evaluate the potential of applying enzymatic interesteri ﬁ cation (EIE) to produce new healthy zero -trans high vitaminic content margarine fats, based on rapeseed oil (RO) only, with a speci ﬁ c focus on process-induced contaminants: the dialkylketones. Three blends made of RO and fully hydrogenated rapeseed oil (60:40, 70:30 and 80:20 wt:wt%) were considered. Compositional and melting properties, polymorphic behavior, appearance and textural characteristics were compared before and after interesteri ﬁ cation. Interesteri ﬁ cation improved both functional and textural characteristics. Enzymatic interesteri ﬁ cation produced fats which presented similar triacylglycerol pro ﬁ le and physicochemical properties as the chemically interesteri ﬁ ed one (CIE), however, preserving more tocopherols and without forming dialkylketones (process-induced contaminants). On the contrary, it was shown that after CIE, dialkylketones were formed: C18:1 – C18:1 and C18:1 – C18 chain-lengths being the major ones, and interestingly, their amount increased with the unsaturation in the blends. Our results indicate the great potential of enzymatic interesteri ﬁ cation to produce new healthy fats entirely based on rapeseed oil. This study provides valuable information for the food industry, showing EIE as cutting-edge sustainable technology.


Introduction
Fat modifications such as fractionation, interesterification or hydrogenation make it possible to modify the fingerprint of an oil in order to broaden its use in the food industry. Fractionation is a physical process based on crystallization; interesterification and hydrogenation are chemical processes using catalysts. In the past, the food industry widely utilized partial hydrogenation to produce base-stocks for margarine formulation. Partial hydrogenation creates, however, high amounts of trans-fatty acids (TFA). The formation of transisomers was previously considered as an advantage because they crystallize faster, have higher melting points and better oxidative stability than cis-isomers. Partial hydrogenation is now banned in most countries since a direct relationship between TFA and increased cardiovascular risks has been recognized (Mozaffarian et al., 2009). A study in Denmark, showed the correlation between the reduction of the TFA content in food, as a consequence of the law that initially limited TFA content, and the lower cardiovascular disease (CVD) mortality (Wilczeka et al., 2017). Partially hydrogenated oils are no longer "generally recognized as safe" (GRAS) by the Food and Drug Administration (FDA), and on May 2018, the World Health Organization called for the elimination of industrially-produced trans-fats from the global food supply by 2023 (WHO, 2018). Fractionation and interesterification processes are now largely admitted as the main industrial alternatives (Gibon, 2012). Interesterification (IE) redistributes the FA on the glycerol by ester-ester exchanges. As a result of these exchanges, IE modifies the triglyceride (TAG) profile without changing the saturation degree, creating new physicochemical properties (Gibon and Kellens, 2014). This technology has great potential for modifying lipids for their use as commercial products and is now commonly used in the food industry (Gibon, 2021). The literature also indicates the great potential of IE to produce trans-free margarine fats from different edible lipid sources (Nekrasov et al., 2020). Some examples have been reported by Zhao et al. (2014), Ruan et al. (2014), Shi et al. (2015), Wirkowska-Wojdyła et al.  Segura and Jachmanián (2020) and Cui et al. (2022). All these reports indicate that the intreseterified fats may potentially be used in margarines preparation, as IE generally broadens the plastic range and modifies the polymorphic stability. The development of lipids with improved nutritional properties is now of increasing interest and in this context, other sources of lipids with essential fatty acids and/or bioactive components worth to be investigated for making versatile products with good technofunctional and nutritional properties by interesterification (Sivakanthan and Madhujith, 2020). It has been proven that polyunsaturated fatty acids, and specifically linoleic (L) and linolenic (Ln) acids, two essential FA, are effective in the prevention and treatment of coronary heart diseases (CHD) (Wijendran and Hayes, 2004). Rapeseed oil (RO), which is now the third most consumed vegetable oil worldwide (Statistica, 2021), has a high oleic (O) acid content and a well-balanced un-saturated FA composition (L:Ln ratio of 2:1) (Gunstone and Gunstone, 2006). In this context, and in the attempt to develop new healthy zero-trans margarine-fats totally based on RO, fully hydrogenated rapeseed oil (FHRO) was selected as hard fat and blended with RO before being interesterified. RO and FHRO have a low palmitic (P) content; FHRO has a very high stearic (S) acid content (>90%) which has no effect on the cholesterol blood level because it is quickly transformed into oleic acid in the liver (Stanley, 2009). Substituting palmitic acid with stearic acid lowers LDLcholesterol concentrations, which is a well-known risk factor for CHD (van Rooijen and Mensink, 2020). IE can be catalyzed by a chemical, such as sodium methoxide (chemical interesterification [CIE]) or by a lipase (enzymatic interesterification [EIE]). CIE has been commonly practiced in the industry for a long time; EIE is now receiving more attention as an emerging technology for the commodity and the specialty fats industry (Zhang et al., 2020). Compared to CIE, EIE offers several advantages: milder conditions, lower oil losses and better oil quality (Gibon and Kellens, 2014;Sivakanthan and Madhujith, 2020).
Moreover, in 2006 CIE was associated with the formation of underinvestigated process-induced contaminants, i.e. dialkylketones (DAK) (Verhe et al., 2006). DAK are a class of foreign substances which are potentially toxic. Literature regarding this class of components remains however very scarce. The possible formation of long-chain DAK as byproducts of the CIE process was described by Verhe et al. (2006), and a formation mechanism during CIE, involving a b-keto-ester was tentatively proposed. Salgarella et al. (2021) recently reported the formation of DAK during deodorization: the DAK content increased with temperature, and the processes appeared to be significant above 240°C. The chemical structure of DAK was elucidated both by GC-MS and LC-HRMS by Santoro et al. (2018). DAK are ketones deriving from FA and having two straight alkyl chains ranging from C10 and C24. The alkyl groups may be the same or different (chain length and unsaturation). A few studies reported the determination and quantification of DAK using GC-FID (Kato and Watanabe, 2022) and further identification using a GC-MS system (Mariani and Bellan, 2011;Mascrez et al., 2021).
Up to now, very little is known about the long-chain DAK, and their potential impact on human health remains unclear; they are generally considered as a health concern. This is the reason why the German Federal Institute for Risk Assessment set a limit for their migration from packaging (BfR, 2017): it should not exceed 5 mg/kg in food stuff. FDA (2018) states that alkylketene dimers and their hydrolysis products (DAK) should not exceed 0.4 wt% of the paper or paperboard. However, so far, no legal limits are set for food due to a lack of both occurrence and toxicological data. Nevertheless, the importance to avoid their formation in food is highlighted on a patent for the process production of infant formula (T'Zand et al., 2009) To the best of our knowledge, there is no report on using the interesterification process on RO-FHRO blends to produce new healthy trans-free fats suitable for margarine production, especially regarding the possible formation of DAK. The present study was thus designed to fill these gaps by evaluating the potential of applying EIE to produce zero-trans and high vitaminic content margarine fats using RO only, with a specific focus on DAK occurrence. The TAG compositions (by RP-HPLC-RI), the solid fat content profiles (by p-NMR), the melting curves (by DSC), the polymorphism (by powder X-ray diffraction) and the hardness (by penetrometry) were examined and compared before and after EIE and CIE. The DAK content was assessed using SPE and GC-FID after microwave-assisted saponification. The DSC melting curves and the TAG compositions were analyzed as a function of the batch-EIE reaction time; they were used to highlight the modifications in melting properties and to quantify their kinetics.

Materials
Refined RO and FHRO were obtained from Cargill (Hamburg, Germany and Izegem, Belgium). Lipozyme TLIM ® , a Thermomyces lanuginosa lipase, immobilized on silica carrier, which rearranges the fatty acids on the glycerol preferentially in a random way, was kindly supplied by Novozymes A/S (Bagsvaerd, Denmark). Three blends were considered, containing 60% (blend-1), 70% (blend-2) and 80% (blend-3) of RO (wt%), blended with FHRO. Commercial cooking and spreadable margarines were purchased from local supermarkets. Their fat phases were used as references. To recover the fat phase, the margarines were first melted at 70°C to ensure complete melting, the upper lipid phase was separated and centrifuged for 15 min at 33 000 g. Anhydrous sodium sulfate was added to trap residual water traces and filtered under vacuum at 70°C.

Batch-chemical interesterification
Before performing the CIE at lab-scale, each blend (FFA < 0.05%, PV < 1.0 meq O 2 /kg) was dried under vacuum (50 mbar, 120°C, 30 min) to reach a water content <100 ppm. 0.5 wt% of sodium-methoxide (Merck) was added as a powder to 500 g of dried oil and CIE was performed for 30 min after the appearance of the characteristic dark "brownish" color at 90°C, under vacuum (25 mbar). The reaction was stopped by adding citric acid as a 20% solution after the vacuum was broken; it was further mixed with the oil for 15 min. Afterwards, 1 wt% activated bleaching earth (Süd-Chemie, Tonsil Optimum 210 FF) was added to remove colored matters and impurities (30 min, 90°C, 25 mbar); the CIE-blends were then vacuum filtered over a Büchner, using Whatman paper filter n o 1 (De Clercq et al., 2012;Danthine et al., 2014a). The CIE-blends were not post-deodorized.

Batch-enzymatic interesterification
Batch-EIE was conducted at a lab-scale using a double jacketed reactor (900 g). Fresh immobilized Lipozyme TLIM ® was added as catalyst. The dry enzyme was first conditioned: de-aerated and dehydrated, as recommended by the supplier (Costales-Rodríguez et al., 2009). Batch-EIE was conducted at 70°C for 8 h, with a load of 4 wt% of lipase. The stirrer was set at 200 rpm to ensure an even distribution of the particles. Sampling (5 mL) was performed every hour. The product was withdrawn from the top of the reactor using preheated (80°C) material (pipette and filter cloth (porosity 100 mm [03-100/44, Sefar Nitex, Switzerland]) to ensure the removal of the enzyme and avoid solidification (Danthine et al., 2014a(Danthine et al., , 2014b. The EIE-blends were not postdeodorized.

FA composition
FA composition was determined by GC-FID according to the AOCS C2 2-66 and Cc 1a-13 methods (AOCS, 2017). Data are reported as averages of two independent measurements.

TAG composition
The TAG composition was analyzed by reverse-phase HPLC-RI based on the official AOCS method Ce 5b-89 (AOCS, 2017). Minor practical adjustments were made in order to improve TAG separation, all in compliance with the above-mentioned official method. The analysis was performed on a Waters system equipped with two stainless steel Nova-Pak C18 columns (4 mm, 3.9 Â 150 mm). The mobile phase was an isocratic mixture of acetone and acetonitrile (62.5:37.5 v:v) with a flow rate of 1.2 mL min À1 ; the injection volume was 20 mL. The samples were dissolved in methanol:chloroform (1:1 v:v); a differential refractometer was utilized for the detection. Peak areas were correlated with the quantities of TAG in the sample. Peaks were integrated with Empower Pro "Apex Track" algorithm; peak areas below 4000 area counts (equivalent to approximately 0.04% of the total peak area) were not taken into account. Data are reported as averages of two independent measurements.

Tocopherols content
The tocopherol content (a-, Yand d-tocopherols) was determined by HPLC-FLD based on the AOCS Ce 8-89 method (AOCS, 2017). The analysis was performed with a Hewlett Packard 1050 Series HPLC equipped with 4 modules (autoinjector, FLD-detector (excitation wavelength = 290 nm and emission wavelength = 330 nm), online degasser and pump), using Alltech Alltima Silica 5 mm (250 mm length Â 4.6 mm internal diameter) non-polar column. Isocratic elution was applied with a flow rate of 1.5 mL min À1 with hexane: isopropanol (99.5:0.5 v:v) as mobile phase. The samples (0.1-0.2 g) were first dissolved in 25 mL of hexane and 20 mL were injected. Peak identification was made by comparison with retention times of calibrated standards. Data are reported as averages of two independent measurements.

Dialkylketones (DAK)
The DAK content was determined using microwaveassisted saponification (MAS) using an Ethos-X microwave system equipped with a high-pressure SR 12 rotor from Milestone Srl (Milan, Italy), followed by a solid-phase extraction (SPE) and GC-FID (Thermo Fischer Trace Ultra 1300) and GC-MS (Shimadzu GCMS-TQ8050 NX,) determination (Mascrez et al., 2021). The chromatographic separation was performed on a 30 m Â 0.25 mm i.d. Â 0.5 mm df SLB-5 ms capillary column (silphenylene polymer, practically equivalent in polarity to poly [5% diphenyl/95% methylsiloxane]) kindly obtained from Millipore-Sigma (Belfonte, PA, USA). Briefly, MAS was performed on 2.5 g of fat added of 25 mL of KOH/EtOH/water 2 M. The unsaponifable extract was purified by a lab-made SPE to isolate DAK from the remaining fat components. Finally, GC-FID for quantification and GC-MS for confirmation purposes were used. Data are reported as averages of two independent measurements.

Melting behavior by pulsed Nuclear Magnetic Resonance (p-NMR)
The solid fat content (SFC) melting profiles were recorded using p-NMR spectrometer (Minispec-mq20, Bruker, Karlsruhe, Germany), according to the IUPAC 2.150 a serial method (1 h at 0°C, 30 min at each temperature) (IUPAC, 1987) and according to the method described by Ribeiro et al. (2009), which consists in a parallel method with a longer period at 0°C (2 h) followed by 1 h at each temperature.

Melting behavior by differential scanning calorimetry (DSC)
The DSC analyses were carried out using a Q-2000 DSC (TA Instruments, New Castle, USA), using hermetic pans. Calibration was made with indium (m.p. 156.6°C; DH = 28.71 J/g) and eicosane (m.p. 36.8°C; DH = 247.4 J/g) standards. Nitrogen was used as purge gas in order to prevent condensation in the cells. An empty pan was used as a reference. The melting profiles were recorded using AOCS cj1-94 method (AOCS, 2017). The integration and peak temperature measurements were performed using the Universal Analysis Software version 4.2 (TA Instruments, New Castle, USA).

Textural properties (cone penetrometry)
Samples were prepared and analyzed as described by Danthine (2012). Briefly, the fats were melted for 15 min at 80°C to erase the thermal memory, poured into plastic vessels (40 g) to produce "shortenings" by statically crystallizing them at À20°C for 30 min, in order to mimic at lab-scale an industrial scraped-surface heat exchanger (Danthine and Deroanne, 2003). Samples were subsequently stored at 15 ± 0.5°C for 48 h. Texture measurements were carried out after this "tempering": samples were first held at 20 ± 0.5°C for 1 h, before measuring them in a controlled temperature cabinet using a SMS TA.XT2i/5 texturometer (Stable Micro System, Surrey, UK). Penetration tests were carried out with a cone probe (P/45C) penetrating the product in its container to a distance of 5 mm at a controlled speed of 0.5 mm/s, trigger force was 0.5 g (Lefébure et al., 2013). The maximum penetration force was recorded, corresponding to the hardness of the samples (Detry et al., 2021). Data are reported as the mean of five replicates.

Evaluation of the degree of interesterification (DI)
The DI was determined based on a comparison between the actual TAG composition (before and after IE) and a theoretical random TAG calculation (RAN). RAN composition was calculated based on the FA composition (% mol) following the method described by Costales-Rodríguez et al. (2009). DI was calculated according to Equation (1).
with A = the TAG that decreased the most during the IE reaction (in this case SSS), B = the TAG that increased the most during the IE reaction (in this case = SOO), NIE = before IE, RAN = random calculated.

Statistics
Data were statistically analyzed using Minitab 17 software (Minitab LLC, State College, Pennsylvania, USA). One-way analysis of variance (ANOVA) was implemented to identify statistically significant differences between means. All assays were conducted in triplicate unless otherwise specified. Means and standard deviations were calculated and differences between means were determined with a significance level of a = 0.05.

Compositional properties
The FA composition of RO, FHRO and the three blends containing respectively 60 (blend-1), 70 (blend-2) and 80% (blend-3) of RO (wt%), expressed as the mean of two replications, are presented in Table 1. For the RO, the predominant FA were oleic (64.1%), linoleic (18.5%) and linolenic (8.5%) types; the TFA content was 0.37%. As expected, the stearic acid content in FHRO was high (92.1%), followed by palmitic acid (5.5%). The total saturated fatty acids content of FHRO was 99.9%, demonstrating that the hydrogenation process was complete (only 0.1% of C18:2cis remained). The average TFA contents were 0.17%, 0.18% and 0.22% for blend-1, À2 and À3 respectively. They are all suitable to produce margarines, as, since 2019, the maximum TFA authorized amount in processed food products is 2% on a fat basis (UE 2019/649, 2019). Due to their high RO content, the blends also had a favorable C18:2n-6/C18:3n-3 ratio of 2.2; this is of considerable importance since the goal was to formulate new healthy foods.
The three blends were both CIE and EIE and compared to NIE materials. As expected, no changes occurred in the FA profiles since IE does not affect the FA composition. On the contrary, analyzing the TAG profile is highly indicative for monitoring the randomization occurring during IE, thanks to the degree of interesterification (DI). The TAG composition of the NIE, the actual TAG composition of the CIE blends and the calculated theoretical random composition (RAN) are presented in Table 2. SSS, OOO, OOL, and LLO were the major TAGs in the NIE blends. The highest SSS content (32.9%) was seen in blend-1 due to the highest FHRO content, while OOO, OOL and LLO contents were directly linked to the amount of RO. As expected, the TAG composition was significantly modified after IE. A DI of 100% was obtained after CIE, indicating complete randomization; CIE blend-1 (CIE-1), -2 (CIE-2) and -3 (CIE-3) TAG compositions closely fit with the RAN corresponding ones (Tab. 2). The TAG diversity increased after IE: randomization resulted in the formation of new TAG species such as SOS, due to the high stearic and oleic contents in the original blends, and POS þ SSL; the percentages of SOO, POL þ SLL þ SOLn and POO þ SOL increased significantly while SSS, OOO, OOL and PSS contents decreased. After 8 h of enzymatic reaction, the TAG composition was identical to CIE, with a DI of 100% for the three blends. Figure 1 summarizes the TAG modifications due to IE. It is obvious that the tri-unsaturated TAG (UUU), like OOO, and the tri-saturated TAG (StStSt) like SSS, were reduced while the di-unsaturated TAGs (StUU), like SOO, and the di-saturated TAGs (StUSt), like SOS, POS and POP, were increased (St = saturated, U = unsaturated). These modifications were positive regarding the targeted food applications. Indeed, being (semi-)solid at ambient temperature but (semi-) liquid at body temperature, StUSt usually improve the structural plasticity and gloss of products, while StUU favor lubricity and spreadability.
Regarding the EIE reaction kinetics, the TAG composition was already drastically modified after one hour of reaction (DI > 94%) for the 3 blends (Tab. 3a-c). The DI approached 100% after 6 h of reaction, and at 8 h reached an identical TAG composition as for CIE.

Differential scanning calorimetry
As shown in Figure 2, the DSC melting profiles of the three NIE blends presented two sharp endothermic peaks: a low melting (LM) peak at about À11°C and a very high melting (VHM) peak at 62/65°C. The LM peak was due to RO and the VHM peak to FHRO. Both CIE and EIE lead to drastic changes in the melting behaviors thanks to the better TAG compatibility. As a result, the LM peak observed for the NIE-blends was shifted to higher temperatures as a broader LM endothermic area (shoulders at 2/À6/À8°C) and a well-resolved medium melting (MM) peak (5/18°C, depending on the blends); contrarily, the VHM was shifted to lower temperatures, as a broader endothermic high melting (HM) peak (49/45/40°C, depending on the blends).       DSC curves of samples withdrawn every hour during the enzymatic reaction are presented in Figure 3a-c. After only one hour, the melting profile changed drastically whatever the blend, as for the TAG composition: the LM peak observed in the NIE-blends was split into two broad areas on either side of the original peak. It was then progressively transformed into a well-resolved MM peak which moved to reach about 18°C for blend-1, 12°C for blend-2 and 5°C for blend-3. This resulted from the decrease of the low melting point UUU-TAG content and the formation of new medium melting point TAG (StUU and StUSt). At the same time, the HM peak was shifted to lower values and reached 49°C for blend-1 and 45°C for blend-2, and about 40°C for blend-3 after 8 h of reaction. This resulted from the decrease of the content in high melting point StStSt and the increase of StUU and StUSt (Fig. 1). The new patterns remained quite similar from 6 h to 8 h of reaction, as it was for the TAG composition (Tab. 3a-c).

Solid fat content
Melting profile and melting point (MP) are parameters of significant importance for characterizing and developing edible fat. SFC greatly influences the suitability of fats for a particular application. The amount of fat crystals is responsible for many product characteristics, including general appearance, ease of packing, organoleptic properties, ease of spreading, and oil exudation (Karabulut et al., 2004). The SFC melting profiles (obtained using IUPAC 2.150 a serial method) of the NIE-blends, the CIE and the EIE-blends, are presented in Figure 4a-c. The SFC profiles of all IE-blends were significantly different from the non-interesterified ones (P < 0.05).
Due to the presence of FHRO, all the NIE were completely melted at about 65°C, with a higher SFC and a steeper melting profile for blend-1 compared to blends-2 and -3. After IE, they were all softer at all temperatures with MP shifted to about 55°C, 50°C and 40°C, respectively; the SFC melting profiles were also steeper at all compositions. This lower MP can be explained by the substantial reduction of the SSS content (whose melting point is 65°C) from 32.9, 24.7 and 16.4% to 4.0, 2.2 and <1.0% respectively. Figure 4a-c shows that CIE and EIE-blends had similar profiles but significantly different from the NIE-blends; it is evident that IE lead to an important reduction of the SFC for the 3 blends, at all temperatures. This was especially true for temperatures higher than 30°C. For example, at 40°C, SFC ranged from about 14% to about 31% for the NIE blends, while, after IE, they decreased to about 1-9% only. It is important to notice that the SFC melting profiles of all the IE blends had an unusual shape with a decrease around 5°C for blend-2 and blend-3 and 10°C for blend-1 (see circles on Fig. 4a-c). This behavior is linked to the new TAG composition, which induced slower crystallization kinetics.   Figure 5a-c. The unusual shape observed at around 5-10°C disappeared by using the modified one (arrows). This demonstrates that a longer tempering might be needed to reach equilibrium, which is typical of StUSt based fats.

Food applications
As the compositions and SFC profiles of CIE and EIE fats were similar, for clarity, only the properties of the EIE-blends are presented and discussed.
After EIE, a broader range of medium melting TAG was formed, as aforementioned (Tab. 3a-c, Fig. 1). This should create an optimal melting temperature range for applications such as margarines. On one hand, SSO and POS þ SSL formation and increase in SOL þ LLS þ POL and SOO contents will endow lubricity, aeration, clarity and gloss to the fat. On the other hand, PSS and SSS usually act as structuring agents and moisture barriers; however, a low amount of SSS is essential to avoid unwanted waxy mouthfeel. SFC melting profiles are often used to determine whether a fat blend is suitable for an application or not. For example, SFC values at temperatures between 4 and 10°C give an idea regarding the ease of spreadability of the product when removed from the fridge; SFC should be lower than 32% at 10°C to ensure good spreadability (Karabulut et al., 2004): EIE-blends-3 (EIE-3) and -2 (EIE-2) fited these requirements. Plastic fats should have not only plasticity and spreadability at low temperature, but also maintain their appearance, without any oil diffusion, at room temperature: SFC at 20-22°C determines this product's stability; SFC of about 10% is essential to prevent oiling off. Finally, SFC at body temperature directly influences sensorial and organoleptic properties of the products. Fat melting should happen in this temperature range to release aroma and avoid a waxy feeling in the mouth, no (or very low) SFC at 40°C is required, as fats exhibit their slip melting point when the SFC is reduced to 5% (Noor Lida and Ali, 1998): this was the case of EIE-2 and EIE-3. SFC melting profiles were further compared to the melting profile of fats extracted from commercial products: spreadable and cooking margarines. The melting profile of EIE-3 resembled the melting profile of the fat phase extracted from the spreadable margarine (Fig. 6). The melting profile of the fat phase extracted from the cooking margarine was intermediate between the melting profiles of EIE-1 and EIE-2: it was similar to the melting profile of a mixture of 50:50 wt% of EIE-1 and EIE-2. These results indicated the great potential of EIE to produce trans-free margarine fats fully based on RO.
Finally, the hardness of the EIE-fats was evaluated by cone penetrometry and compared to the same commercial references (Tab. 4). It is obvious that all the EIE-fats were softer than the corresponding NIE ones. The hardness of the reference cooking margarine was very close to a 50:50% blend of EIE-1 and EIE-2 while the spread was very close to EIE-3. The macroscopic appearance of the blends was also modified after EIE (Fig. 7). NIE-blends had a hard and brittle structure,  with an irregular surface and it can be noticed that they were crystallized under the b-form, with a double packing (2 L) (Fig. 8). On the contrary, the EIE-blends, mostly crystallized under b'-2 L-form, with a small proportion of b-2 L-form (Fig. 8). EIE of the investigated blends not only reduced the content of the high-melting point TAG but also significantly increased the number of different TAGs, which in turn, favored crystallization into a b'-form, which is the most desired form for margarines. The new trans-free fats obtained by EIE were more homogeneous, plastic and glossy compared to the initial blends (Fig. 7) and are therefore suitable for margarines production.

Minor components
Besides changing the TAG composition of fats and oils, IE processes can impact the content in minor components. Tocopherols are minor components naturally present in the oil as different isomers: a, b, g and d tocopherols. They have vitaminic activities (a > b > g > d) and antioxidative properties (d > g > b > a), so that their content should be preserved as much as possible. Tocopherol content of RO is high, gand aforms being the dominant ones (Piravi-Vanak, 2020).
Results obtained for blend-2 showed that the total tocopherol content was highly affected by CIE, which led to a significantly stronger reduction of the total tocopherol content compared to EIE (Tab. 5). This is consistent with data reported by Costales-Rodríguez et al. (2009) for CIE of a blend of palm stearin-soybean oil. The vitaminic activity was better preserved thanks to the milder conditions of the enzymatic process. Table 6 shows that CIE produced potentially toxic DAK, while EIE did not. Regarding the type and amount formed during the CIE, C18:1-C18:1 and C18:1-C18 were dominant in the three CIE-blends, followed by C18-C18 and C18:1-C16. This was obviously linked to the FA composition of the initial blends as both C18:1 and C18 were the major FA (Tab. 1). Moreover, interestingly, the type and amount of DAK formed during CIE depended on the initial fat composition, and especially the unsaturation degree: the highest amount was found in blend-3 (422 mg/kg), then blend-2 (128 mg/kg), then blend-1 (117 mg/kg), meaning that the amount of DAK formed decreased with the saturation degree. Blend-3, which was the most unsaturated, contained the highest amount of C18:1-C18:1, C18:1-C18 and C18:1-C16 (Fig. 9).

Conclusion
This study showed that using FHRO as hard fat and RO as soft oil is a suitable alternative to produce high-quality zero-trans margarines by interesterification. The optimal omega 3/omega 6 ratio of 2.2 is retained in the interesterified products. The interesterified blends have melting characteristics and hardness scales that can simulate commercial margarines, thus IE of RO and FHRO can potentially be used to produce high-quality zero-trans fats. However, the triglyceridic composition of the new fats is such that their crystallization rate is slowed down. Regarding DAK formation during CIE, it was shown that the type and amount depended on the initial fat composition, and especially the unsaturation degree. Our results indicate the great potential of IE to produce new fats fully based on RO, however, EIE is preferred to CIE for the preparation of healthy trans-free margarine fats for two reasons: (i) better preservation of tocopherols: vitamin supply and preserved oxidative stability and (ii) absence of processinduced contaminants (i.e., DAK). As a consequence EIE can be viewed as cutting-edge sustainable technology for the food industry.