A novel strategy for foodstuff can coatings: Hybrid resinous materials beside silica nanoparticles

Abstract For many years, Bisphenol A (BPA) based epoxy resins have been used as coating material for foodstuff cans because of their appropriate properties. Recently, with attention to the harmful health effects of BPA derivatives, limitation of the usage of epoxy resins has escalated. In this research, using Taguchi experimental design method, different nanocomposite coatings based on hybrid resinous materials containing different amounts of epoxy and polyester and silica nanoparticles were formulated. The migration properties of formulated coatings were investigated based on the standard methods. The results showed that the 45 wt.% replacement of epoxy by polyester resin in combination with 2 wt. % of nano silica, had not any inverse effect on needed mechanical, physical and chemical properties of optimized formulation and, at the same time, migration of BPA and toxic materials were in the acceptable ranges, based on the EN1186-EU 10:2011 standard. Graphical Abstract


Introduction
These days, metal-based packaging is used extensively in food industries due to their outstanding barrier properties. Metals potentially react with aggressive food products, and consequently, corrosion will happen. Protective coatings are needed to barricade the metal-food contact, the subsequent interactions, corrosion and contamination of food products by corrosion products, leakage of the corroded cans, and more importantly, the migration of metal compounds into the foodstuffs. [1,2] Traditionally metallic cans have been protected against corrosion by polymeric materials, especially bisphenol A (BPA) based epoxy coatings, as an inner barrier between food and metal in cans. [3] Epoxy-based resins are one of the best choices for applying them on inner surfaces of cans due to their excellent chemical resistance, compatibility with more food and beverage types, and appropriate adherence to the metal substrates. However, in the case of insufficient polymerization conditions, the BPA as a potentially toxic compound having estrogenic activity could be present in the thin inner coating. Hence, the migration of unreacted BPA toward the foodstuffs poses a serious food safety concern toward consumer health as the first priority in this application. [3][4][5] Materials that come into contact with food in each stage of production, process, packaging, and preparation are defined as food contact materials. [6] Migration of the components of these substances to food may affect the organoleptic properties of food and, if not adequately controlled and in amounts more significant than the specified toxicity, have detrimental effects on health. Restrictions on chemical additives and their migrations are fully established in accordance with international regulations. [7] According to (EU) 2018/213, [8] specific migration limit (SML), which is the allowed amount of BPA to migrate into the food, Due to new EU regulations and the fact that non-food sources of bisphenol A also play a role in general exposure, the SML level was reduced to 50 micrograms per kilogram for plastics in 2018. This restriction has been valid from September 6, 2018, and applies to varnishes and coatings in contact with food. The Migration of BPA and its derivatives is usually controlled by using food simulants such as water, acetic acid, aqueous ethanol, and edible oils.
An effective solution to resolve the problems caused by BPA derivatives is to develop the replacement for epoxy resin in can coatings. Different candidates such as acrylic and polyester resins are the most common materials which were proposed for replacement of epoxy coatings. [9,10] Also, other types of polymer coatings like polyolefin and non-BPA Epoxy coatings have been developed. [11] Generally, acrylic resins possess outstanding corrosion and sulfide stain resistance. However, the FDA approved grades of acrylic resins for applications with food contact have less corrosion resistance. On the other hand, acrylics also are too brittle to resist in the fabrication process of drawn cans.
Polyester resins, which are the product of carboxylic acids and alcohols, are another group of the proposed replacement. They could be easily applied to the fabrication process because of their appropriate flexibility. [5,10,12,13] Polyester as can coating does not change the taste or odor of the foods and beverages. [12] Although, polyester can coatings are less resistant to corrosion than epoxy coatings, they are noticeably safer alternatives. The polyester resins used in can coating should be produced from the material, especially catalysts, [8,14,15] which have dietary allowance (FDA) [13] CFR175/300. [16] It is necessary to analyze the migration of compounds from packaging materials into the foodstuffs by using different analytical methods, such as gas or liquid chromatography, UV absorption. [7,17] Nanotechnology has found its place in food contact materials in the last decades owing to the potential of nano additives to improve some properties of materials or to upgrade them decently. [18,19] As SiO 2 nanoparticles incline to fill pinholes created through the curing process and also the subsequent shrinkage, they act as bridges in the interconnected matrix. As a result, nanoparticles modifying epoxy coatings display improved anti-corrosion and mechanical properties, simultaneously. [17] Besides decreasing the free spaces by the presence of nanoparticles, they could play a vital role in activating sites that increase the cross-linking density of the resultant coatings. [18,[20][21][22] This work aimed to introduce a new formulation of can coating having a less amount of epoxy resin to minimize the BPA migration into the food ingredients by replacing considerable amounts of epoxy resin with polyester using appropriate experimental design methods. For this purpose, pure epoxy resin, polyester/epoxy, and polyester/epoxy-containing nano silica were formulated, considering to the availability of raw materials and economic issues. The coating samples were examined mechanically, chemically, morphologically, toxicity, global migration, and BPA migration. On the other hand, for evaluation of thermal degradation trend of formulated coatings, which are subjected to thermal processes in food industries like pasteurization or sterilization, thermo gravimetric analysis (TGA) was performed.

Materials
Polyester resin coded as DUROFTAL PE 6160/50MPAC and melamine formaldehyde resin coded as MAPRENAL MF 904/97 were purchased from CYTEC Co. (Germany). Bisphenol-A based epoxy resin (KUKDO YD 019) was supplied by Kukdo Chemical (South Korea). Modified SiO2 nanoparticles, (AEROSIL R972) as a hydrophobic nano SiO 2 with a specific surface area of 90-130 m 2 /g, the content of >99.8%, tamped density almost 50 g/l and pH equal to 3.6-5.5 was purchased from Evonik Industries (Germany). All the used resins, as well as nano SiO 2, were food grade. BYK CATALYST 450 accelerator for cross-linking and BYK 310 as a heat resistant and surface property improving agent, and Disperbyk 161 as a dispersing agent for nano SiO 2 were obtained from BYK Co. (Germany). A solvent including methoxy propyl acetate (99.90%) was used (METRON, France). Titanium dioxide grade 2971 was obtained from KRONOS (United States). Blocked aliphatic polyisocyanate known as DesmodurV R VP LS 2078/2 were purchased from Covestro Industries (Germany). All of the chemicals were used without further purification. The raw materials which were mentioned in Table 1, were used to make food simulation solutions.

Design of experiments
The method for selection of the best formulation was based on the Taguchi experimental design method using Qualitec4 software regarding the number of variable and level. According to the Taguchi method, a standard L 9 orthogonal array was selected ( Table 2). Based on the accepted approach in applying different runs to eliminate the probable systematic bias, randomized selection of runs was carried out. [23] The variables in this study were the amount of epoxy resin, nano-silica as well as exact amounts of melamine and catalyst as effective factors in the quality of coating ( Table 3). The final formulation was adjusted in two steps.
Step 1: 9 samples were selected by the Taguchi method, and the best one was chosen by considering the highest chemical and mechanical resistances. (In the conclusion and discussion section, the first test of these samples is listed in Table 1S (supporting information). Finally, sample 7, named S 7 , was selected as the best formulation based on mechanical and chemical tests.
Step 2: Because of some drawbacks in the best selected combination of the first stage, in this step, with modification of formulation (reduce the amount of melamine, increased polyester resin and slight of epoxy resin to improve properties), the final formula named S 7 *, was attained. For comparing results and performances of 4 formulations consisting of S 7 , S 7 *, S 7 * without Nano SiO2 and S neat Epoxy , the mentioned samples were evaluated carefully (Table 4).
Raw materials, including polyester/epoxy resins, solvents, nano SiO 2 , titanium dioxide and dispersant were mixed for 60 min at 3000 rpm. Afterwards, the mixtures were grounded for 120 min in milling apparatus using zirconium balls having a diameter of 1 mm. Assuring attainment of qualified dispersion in the combination, the melamine and additives were blended firstly for 45 min at 1500 rpm.
Tinplate sheets with the dimension of 55 Â 75 Â 0.2 mm were chosen based on ASTM A623M. The sheets were washed with methyl ethyl ketone to remove any contaminant or residual oil. Then uniform wet films having a thickness of 23 ± 5 mm (similar to the usual can coatings) were applied on the tinplate sheets uniformly by using a pen applicator. After 5 to 7 min of flash off, the samples were cured at 190 ± 1 C for 17 ± 1min.

Characterization and test methods
Fourier transform infrared (FTIR) spectroscopy. Fourier transform infrared (FTIR) spectra of the formulations, as KBr pellet containing soft powder of cured samples and also an uncured resin layer with a low thickness on the KBr pellet were depicted via a Bruker instrument, model IFS48 spectrophotometer.
Thermo gravimetric analysis (TGA). Thermo-gravimetric analyses (TGA) of cured coatings samples were evaluated by utilizing TGA-PL, model TGA1500 (Mettler Toldo Co.) apparatus under an atmosphere of nitrogen from ambient temperature to 800 C at a heating rate of 10 C/min.
Scanning electron microscopy (SEM). The optimum samples surface morphology and particles distributions of nanoparticles were focused an applying Scanning Electron Microscopy (SEM). The used tool was QUANTA 200 system (FEI Company, USA). A very thin layer of gold was sputtercoated to improve the surface conductivity by the PVD method in a COXEM instrument (South Korea). The EDAX map was also obtained to detect the distribution quality of the nanoparticles in the resin matrix.
Transmission electron microscopy (TEM). The transmission electron microscope, Philips EM 208S (Netherland), with accelerating voltage of 100-kV, RESOLUTION: 0.3, was employed to study the morphology of the prepared nano composite can coatings. Ultra-microtome at À55 C was utilized to achieve the ultrathin slides (thinner than 80 nm) cuttings cryogenically due to employing a diamond knife manufactured by a Leica Ultra cut (UCT, Germany).
Investigation of mechanical properties. Adhesion of the coatings to the substrate was investigated using a Gardner cross-cut adhesion tester (BYK, Germany) according to ISO 2409:2020. Accordingly, after magnification of 6 to 11times, the dry film adhesion on the tin plate sheets was evaluated. Impact test (Direct, indirect) was performed to evaluate the stability of dry film on the metal sheet with dimensions of 70 Â 70 Â 0.2 mm the used tool was impact tester QCM (Sheen, UK) and tests were performed according to DIN EN ISO 6272.
Coating's hardness test was performed to determine the scratch resistance of the dry films using a hardness meter DUR-O-Test (BYK GARDNER, Germany) having 1 mm diameter of the needle according to ISO1518-1.
Chemical resistance test. Chemical resistance test was performed to evaluate the resistance coatings against pasteurization and sterilization conditions in an autoclave at standard temperature, pressure, time, and food simulation solutions. The standard solutions were consisted based on Iranian Standard INSO 2509. This standard is specified for organic coatings evaluation used in packaging industries.
Five simulant solutions were prepared as followings: Solution 1: 1.9 ml of glacial acetic acid and 3 g of NaCl were mixed and the volume was increased to 100 ml using distilled water. Solution 2: 1.1 ml lactic acid and 2 g NaCl were mixed and the volume was increased to 100 ml using distilled water. Solution 3: 2 g citric acid monohydrate and 3 g NaCl were mixed and the volume was increased to 100 ml using distilled water. Solution 4: 0.5 g L-Cysteine hydrochloride monohydrate was completely dissolved in a phosphate buffer solution having pH 7. Then its volume was increased to1000 ml using distilled water. This solution should be prepared at least 4h before the test, or it should be heated to 70 C and used after cooling (This solution is used for the sterilization process). Solution 5: 2 g citric acid monohydrate and 10 g of sugar were mixed and the volume was increased to 100 ml using distilled water.
After pasteurization and sterilization of coated plates in the autoclave, they were observed for any sign of cloudiness (compared with the control sample), having black or brown spots due to tin dissolution, pealing, scaling, corrosion or appearance of red-gray spots, softening and blisters.
Experimental study of global migration and inductively coupled plasma optical emission spectroscopy (ICP-OES).
The global migration test, as well as heavy metals migration (ICP toxicity test) including Pb, Cr, Cd, Co, and Si, was carried out in a solution of 20% (v/v) ethanol and 3% (v/v) acetic acid as a simulant of food mediums [24] according to BS EN 1186 and Directive 94/62/EC test methods, respectively. For this purpose, the prepared samples were put in metal cells containing the simulants at 40 C in a temperature-controlled cabinet for 420 h. afterward; the aliquots of each migration solution were calculated according to the above-mentioned standards.
BPA special migration test. To determine the migration of BPA into the various stimulant media, High-performance liquid chromatography (HPLC) test was used after exposure of coatings in simulants. This test of solutions was used to determine the amount of BPA migration from different coating samples to different simulant environments. Highperformance liquid chromatography device, model 1200, (AGILENT, USA) along with a fluorescent detector, an excitation wavelength of 225 nm, an emission wavelength of 305 nm. Column with specifications Agilent Eclipse XEB C18, It is 4.5 Ã 250 mm in length and diameter, with a particle size of 5 lm and a flow ratio of 1-1.2.
The mobile phase is Acetonitrile: Water, with a gradient of 45:55 to 65:35 and a column with a temperature of 30 C. The time to appear of each sample was 17 min and the amount of injection was 10 mL. The concentration of each sample was measured with three replications.
For evaluation of specific migration, the coated plates were placed into the glass containers containing various types of simulant solutions Table 1S (supporting information). According to the EN 13130-1 standard, the conditioning included applying a temperature of 100 C for one hour (pasteurization stage) and then for 10 days at a temperature of 40 C (durability). After ending the test time, the glass containers were removed from the autoclave, and after cooling, the solutions were examined for BPA specific migration.

Results and discussions
According to Table 3, the 9 samples were proposed by Qualitek-4 software were chosen to evaluate their characteristics by cross-cut, impact, hardness and chemical resistance tests in the results are shown in Table 5. The hardness test is one of the most essential tests in food cans. Qualitek-4 software has been utilized for circumstances optimization and ANOVA analysis to determining of contribution of each factor ( Table 6).
The percent of the efficiency of each factor was obtained by the ANOVA method is given in Table 6 the importance of the factors based on ANOVA data was as following: (1) Amount of Epoxy resin in the formulations, (2) Nano SiO 2 content, (3) Melamine resin content and (4) amount of catalyst. Table 5 also shows that the S 7 sample was the best formulation among 9 different samples due to its relatively good properties in chemical, hardness and impact tests and has similar results to S neat Epoxy . Hence, one further modification step was essential to perform on formulation [7] (S 7 ) to obtain a more decent coating. The critical role of epoxy ratio in the formulation, especially for hardness resistance, was proved by the ANOVA software ( Table 5). As a result, the final optimization phase was done by addition of Nano silica into the formulation S 7 leading to the creation of formula S 7 *, (as depicted in Table 4).
Adjusting the amount of epoxy in formula S 7 * was accomplished delicately via considering two concerns: 1) control over immigration of Bisphenol A and 2) reinforcing the coating structure for tolerating chemical attacks and mechanical tensions. However, the coating material based on formula S 7 * could pass the tests successfully with acceptable improvements.

Characterization of FT-IR
FTIR was applied to identify the resin curing process and the probable reactions included in it. Figure 1 shows the overall spectra of the prepared cured coatings with and without nano SiO 2 , the decrease of the peak at around 825-918 cm À1 related to oxirane ring breathing [25] for the crosslinked resin with and without nano SiO 2 , as well as intensity reduction of a broad peak at 3445 cm À1 (belongs to hydroxyl functional groups of hydroxyl ended polyester resin and Epoxy resin) approves the formation of 3-dimensional hybrid network. After the curing of the sample, this peak was shifted to 3290 cm À1 for blended resin samples with and without nano SiO 2.
[22] Reduction of oxirane peak intensity indicating the reaction of polyester resin hydroxyl attack the in DGEBA resin epoxy part.
Meanwhile, the appearing peaks at the region of 2900 cm À1 have corresponded to the C-H groups existence for the diglycidylether of bisphenol A (DGEBA) of epoxy resin. [23] The spectrum of pure Epoxy resin was included the typical absorption bands attributed to a bisphenol A and epichlorohydrin. In S neat Epoxy sample after curing OH bond obviously appears between 3600 and 3200 cm À1 represents absorption characteristic of the obtained band. Nevertheless, the low intensity and short form for the peak   Figure 1. FTIR spectra of the S 7 Ã before curing, S 7 Ã without Nano SiO2 after curing, S 7 Ã after curing, S neat Epoxy after curing, S neat Epoxy before curing samples.
can be interpreted as a low number of hydroxyl functional groups in the chemical structure. According to Scheme1, there are two possible hydroxyl groups to react with methoxy groups of melamine resin; first, hydroxyl functionalities of the polyester and second, hydroxyls in the middle of a DGEBA Epoxy molecule. On the other hand, the OH groups at the end of a polyester chain can participate in a ring-opening reaction. So, the ring-opening mechanism of the epoxy curing process in the structure will create secondary hydroxyl groups. Consequently, a shift to the lower wavenumber for the broad peak of hydroxyl is reasonable. The broad peak at 3445 cm À1 for the blend sample could be associated with intermolecular hydrogen-bonded of the hydroxyl group in the polyester chains and polyester with epoxy molecules. [26,27] After the curing reactions, the nature of these interactions changed intrinsically.
It should be noted that the intensity of the peaks attributed to C-N stretching at 1556 cm À1 , and the 1378 cm À1 represents the C-H bending of -N-CH 2 OCH 3 group for melamine molecules are slightly decreased after curing. [28] Figure 1 also shows that the presence of nano SiO 2 has not any special effects on the curing reactions of the coating samples, and the characteristic peaks of S 7 * without Nano SiO2 and S 7 * do not show any significant differences.
The peak at 937 cm À1 was the stretching vibration of the Si-OH stretching peak. It indicated that nano-silica existed in the cured S 7 * . [29] Effect of the presence of SiO 2 nanoparticles on the morphology of polyester/epoxy coatings Typical SEM images of the optimum coating S 7 * (polyester/Epoxy, containing nano SiO 2 ) are shown in Figure 2. Figure 2a shows that the coatings modified by SiO 2 nanoparticles have a relatively homogeneous morphology. Due to the high specific surface area of SiO 2 nanoparticles and subsequently their ability to release internal stress and tension to the resin matrix, there is no sign of nanoparticle agglomeration. Speculated that the higher specific surface area of nanoparticles leads to the better dispersion of nanofillers into the matrix and more actively nuclei for the growth of cross-linking networks in the curing process. It has been claimed that the very small size of nanoparticles enables them to be penetrated the even super-small cavities of the coating matrix as well as the metal substrate. The mentioned mechanism can block holes and paths to stop crossing any external and the internal materials from immigration to the next phase. Hence, it can enhance barrier property of the coating, which is one of the primary goals in can coatings. [30] The homogeneous and integrated morphology of the coatings indicates the uniform dispersion of nano-SiO 2 particles within the resin matrix. This could be advantageous for the final mechanical strength as well as the barrier properties, [31] which is indirect and long-term contact with food products and beverages.
Good dispersion of nanoparticles can be attributed to their spherical shape and considerable interaction between dimethyledichlorosilane (DDS) functionalities on the SiO2 surfaces and the Epoxy/polyester organic base causing the formation of hydrogen bonds with higher density. The  incorporation of organic compounds into SiO 2 particles promotes hydrophobicity feature of them leading to attain higher levels of compatibility between the components of the coating, and consequently, it facilitates the diffusion of SiO 2 flakes into polymeric chains.
The absence of agglomeration and the uniformity of nano SiO 2 distribution have been confirmed by the SEM-EDX map of the film surfaces (see Figure 3).
Typical TEM images of the optimum coating S 7 * (polyester/Epoxy, containing nano SiO 2 ) are shown in Figure 4 TEM results illustrated that the nanoparticles distribution in the polymer matrix was homogeneous. The particles have been dispersed spherically throughout the sample with nanometer range dimensions. The uniform dispersion of the bright regions into the polymer matrix signifies the homogeneous dispersion of nano SiO 2 particles. As shown in Figure  4, no SiO 2 particle aggregation is found, it showed that appeared nanoparticles well-dispersed in the resin matrix regions caused by the settlement of some functional groups upon the nano SiO 2 surface such as Dimethyledichlorosilane (DDS) functionalities on the SiO 2 surfaces upon the nano SiO 2. [32,33] Thermo gravimetrical analysis (TGA) The TGA and DTG thermograms of the prepared coating films (S neat Epoxy , S 7 * without Nano SiO2 , and S 7 *) have been represented in Figure 5 (a,b). The main aims for this test were: (a) considering the coating thermal degradation changes affected by replacement of about 47% of DGEBA Epoxy resin with polyester and (b) study of thermal  degradation trends after addition of nano SiO 2 to the coating formulation.
As shown in Figure 5, in the neat Epoxy-based coating, a single step degradation at 415 C has been occurred. On the other hand, in samples containing polyesters (S 7 * without Nano SiO2 , S 7 *), there are two stages of degradation. The second stage is at about 415 C, which are defiantly related to the DGEBA Epoxy resin of the coating formulations.
As shown in Figure 5, in S 7 * without Nano SiO2 and S 7 * samples, the first stage of degradation, which was appeared at 378 C, can be related to the degradation of polyester portions. As the amount of nano SiO 2 was about 0.5 wt. % in S 7 * formulation, the degradation trend of this sample is somewhat similar to the S 7 * without Nano SiO2 . It could be deduced that, although the thermal resistance of the epoxy coating was decreased by replacing approximately 40% of resin by polyester, the onset temperature of degradation for all of the hybrid coatings are much higher than the sterilization temperatures (It is usually between 105 -121 C depending on the type of product). So, the modified coating formulations had enough stability in the sterilization conditions. Table 2S (Supporting information) also show that they remained char of S neat Epoxy , S 7 * without Nano SiO2 and, S 7 * samples are about 42, 31, and 32 wt. %, respectively. Indeed, the introduction of the aliphatic structure of polyester to the coating formulations have resulted in a decrease in char yields (S 7 * without Nano SiO2 and S 7 * coating samples). [23,34] It should be noted that the small peaks of the DTG thermograms at 217 C for all of the samples could be due to  the physically adsorbed water from the environment, which contributes about 3-5% 23. Also, as shown in Table 2S (Supporting information), the T onst of Epoxy coating (S neat Epoxy ) is less than those of S 7 * without Nano SiO2 and S 7 *. Supposing that the initial weight loss of the samples is related to the release of trapped water in the samples. [35,36] Also, it could be determined that this release happens at a higher temperature for coating containing polyester. Indeed, the polyester containing compounds (S 7 * without Nano SiO2 , S 7 *) have more ability to form hydrogen bonding with water molecules and these bonds were not destroyed by heating. [37] As a result, the onset temperature of S 7 * without Nano SiO2 and S 7 * were higher than that of S neat Epoxy . On the other hands, the temperature that the 10% weight loss is observed is rather similar for all the samples. T 50% (temperature of 50% weight loss) for samples containing polyester is lower than that of epoxy ones. According to these results, as well as the char yield of samples, it could be deduced that the thermal stability of the epoxy coating is greater than that of polyester coatings, especially at higher temperatures.

Mechanical properties and chemical resistance of coatings
Mechanical properties, such as adhesion to the substrate and impact resistance and hardness, are essential factors in can coatings. Results of mechanical properties of prepared coating samples have been demonstrated in Table 3S (Supporting information).
According to Table 3S (Supporting information), S 7 * without Nano SiO2 sample, shows rather similar mechanical properties to the epoxy-based coatings. However, the hardness of S 7 * without Nano SiO2 is lower than that of S neat Epoxy sample and the addition of nano SiO 2 particles has compensated the weak hardness of S 7 *.
All the three samples were placed in pasteurization and sterilization conditions. As shown in Table 3S (Supporting information), it is obvious that, chemical resistance of S 7 * without Nano SiO2 is acceptable and the coating sample containing polyester resin has not passed after exposure in the Citric Acid and NaCl (solution 3). The good chemical resistance of S 7 * compared to S 7 * without Nano SiO2 , could be due to the uniform dispersion of nanoparticles and consequently higher chemical resistance as well as mechanical properties. [38] The addition of nanoparticles into the formulation of a coating can donate significant chemical resistance against acid, alkaline and salt solutions. [28,39,40] This chemical resistance behavior can be justified by Van der Waals and hydrogen bonds between the organic base and nano-silica particles. [38,41,42] It is proved that the adhesion of epoxy coating can be highly raised via the introduction of nano SiO 2 into the epoxy matrix and blocking all pathways as physical hindrance versus corrosive ions and acidic or alkaline attackers. [19,43] Global migration and ICP test Migration tests were carried out to determine the global migration rate from the whole coating into the food or food simulants. The immigration test has complied with international standard EN1186-EU 10: 2011. [44] According to the mentioned standard, the total migration rate of the coating or final product should not be exceeded by 10 mg/dm2 or 60 mg/kg (60 ppm) of food or simulant.
According to the Figure 6, all of the samples, i.e. The Polyester/Epoxy resin containing nanoparticles (S 7 *), the control sample (S 7 * without Nano SiO2 ), and the epoxy coating sample S neat Epoxy, in acidic and alcoholic media, represent an acceptable result in global migration test. The rates of migration in the acidic environment of S neat Epoxy and S 7 * without Nano SiO2 are higher than those of S 7 *. It is also observed that the rate of global migration in an alcoholic environment in S 7 * is much less than that of the other two samples. It could be concluded that the S 7 * can be a better candidate, conventional with epoxy coatings, for using in alcoholic and acidic environments. [45] ICP test was performed to detect the possible impurities in the prepared coatings as a result of residual catalysts as well as the possibility of heavy metals and Si penetration from the (a) tinplate substrate into the coatings, and food simulants. The results are given in Figure 7.
According to the Figure 7, the amount of lead and chrome in the epoxy sample is higher than that of the two other samples (polyester/Epoxy with and without silica nanoparticles). The toxicity of silica particles is also within the allowable range defined by the US Food, Drug and cosmetic Administration regulation 21 CFR 175.300 21 Altogether, all heavy metals and Si must be less than 100 ppm, which results show that all samples are within the acceptable range. [46] Besides the Pb, Cr, and Si toxicity investigation, this test was performed for Cadmium, Arsenic, and Cobalt. All three samples showed no trace of these metals.
The insignificant metals detected can be related to the residual catalyst in the resin or the dissolved metals from the synthesis equipment such as the reactor or mixer. The Figure 8. The Specific migration of Epoxy-based coating (S neat Epoxy ) and Epoxy/polyester-based coating containing nanoSiO 2 (S 7 *).
source of Si in S neat Epoxy and S 7 * without Nano SiO2 samples may be impurities in resins, catalysts in making resins and may come out of the tin substrate.

BPA specific migration to food simulant environments
Migration tests were carried out to determine the specific migration rate from the whole coating into the food or food simulants. The immigration test has complied with international Commission Regulation (EU) 2018/213. According to the mentioned standard, the specific migration rate of BPA through the coating or product should not be exceeded by 0.05 ppm (50 ug/kg, ppb) into the food or simulant, [47,48] Iranian Standard INSO 2509.
As shown in Figure 8, BPA migration rates vary for simulant. Also, in the presence of polyester components, the amount of BPA released in all samples at various simulators has been drastically reduced.
According to the Figure 8, in all food simulant solutions, the S7 Ã sample showed less migration quantity than the other samples. Also, this sample's migration is within the permissible limit. As can be seen, the migration rate is higher in the alcoholic solution, which can be justified considering the solubility parameter Table 4S (Supporting information) of BPA, solvents, and resins.
In general, Solubility is another important parameter. The expected NIAS needs to be soluble enough in the chosen extraction solvent or food simulant, BPA is not well soluble in water and acidic aqueous solutions and can be solved in alcoholic and alkaline solutions. [49] As shown in Table 4S (Supporting information), the solubility parameter of bisphenol A is close to epoxy and therefore, it tends to remain in the epoxy coating and the migration values in different solutions are very small (in the mg/kg range). On the other hand, the amount of BPA migration into the acetic acid, which has the closest solubility parameter to BPA, is much higher than the other solutions. Also, due to the presence of polyester resin in the composition of modified coating formulation (S 7 *), the amount of BPA migration of this coating is much less than that the others. This indicates an improvement in the health and environmental properties of the modified coating. [37,50] Conclusion Epoxy resins have been used traditionally as can coatings, but the BPA can penetrate in to the foodstuffs and pose many health hazards. Therefore, replacing epoxy resins with other counterparts while preserving the chemical and mechanical properties and longevity of the coating, can be an outstanding solution. Accordingly, with helping the experimental design methods, a significant portion of BPA based epoxy resin was replaced by polyester resin. Beside, 0.5 wt. % of nano SiO 2 was added to the hybrid coating formulation to improve the coating properties.
The presence of polyester in modified formulations, did not show any significant descend in the mechanical and chemical properties of the resultant coating. On the other hand, the sample containing uniformly dispersed nano SiO 2 particles, demonstrated rather similar properties to the epoxy coating. All of the modified formulations passed the global migration and ICP tests, and no sign of nanoparticle release into the food simulant were observed and the specific migration of BPA was less than the permissible limit (50 micrograms per kilogram of foodstuff simulants).
Considering all the above results and the existed concerns in terms of BPA problems for human health as well as environmental hazards, it can be concluded that, although epoxy resins are among the highest quality resins, it can be replaced by other biocompatible BPA-free resins such as food-grade polyesters resins. In many cases, the addition of SiO 2 nanoparticles can improve properties of the modified coatings and bring the final coating properties equal to the epoxy resins or even higher.