Food packaging simulant failure mechanisms in next generation steel packaging

Legislative pressures have led to the mature dominant chromium‐coated steel (ECCS) substrate and epoxy phenolic lacquer replacement in Europe. An investigation was carried out to examine the interaction between a steel surface engineered with a novel, developmental substrate coated using Cr (III)‐based electrolytes and the food stuff being canned. Samples of lacquered material were subjected to a typical retort process (121°C for 90 minutes) and examined using a variety of laboratory analytical techniques. The foodstuff being packaged has a significant impact on the substrate/lacquer adhesion with clear differences in failure mechanisms between foodstuffs. There is clear evidence of chemical species transfer through the next generation lacquer, and this can instigate corrosion at the surface where incomplete chromium coverage leads to exposed iron. In general, the novel developmental material exhibits lower adhesive properties and shows a greater sensitivity to the foodstuff, although this is largely attributed to the homogeneity of the coverage. The novel substrate proves to be a promising alternative to ECCS due to REACH legislation, but improvement is required to achieve equivalent performance.

of the metal/organic interface, particularly when subjected to the extreme environment experienced by the packaging when the foodstuff is cooked in the filled can at temperatures in excess of 120°C. 7,8 The two coated steels primarily used for food can applications are tinplate, having a tin layer approximately 0.2 to 2 μm thick and electrochromium-coated steel (ECCS) which has a 10 to 30-nm layer of chromium/chromium oxide, produced using a Cr (VI) electroplating process. 9 Trivalent chromium-based metallic coatings for steel developed 10 as a REACH 6 compliant alternative to ECCS is under iterative development to optimise surface chemistry and characteristics. A novel developmental substrate, combined with next generation coatings (bisphenol-A non-intent) that comply with legislative moves in some European territories, presents the metal packaging industry with a system that has been shown to be less robust than the wellestablished ECCS/epoxyphenolic coating system. 11 The adhesion of an organic coating to a metallic surface occurs through a variety of bonding mechanisms, including electrostatic interactions, dispersion forces, and covalent bonds, 12,13 all of which are susceptible to modification through the interaction with a simulant solution and by extension, foodstuff. Work has previously been conducted studying the effects of the sterilisation process on the quality of adhesion that is observed, a phenomenon that had not been quantified before. 11 This study focussed on the quality of adhesion varied with adjusting retort parameters, establishing relationships and highlighting strengths and failures within novel substrate/coating systems. Previous work has begun the examination process of the changing chemistry of both the coatings and the substrate, an area that is to be more fully explored here. The effect of the retort process on polymer coatings has been subject of some limited study. For example, Axelson-Larsson examined the oxygen permeability of such coatings under autoclave conditions, concluding that the conditions can affect the barrier of some materials, with water acting as a plasticiser. 14 Many studies have been carried out to investigate the effect that the chemical composition of a given foodstuff might have on the coating. 15 The migration study carried out by Errico et al where the BPA transferred into the foodstuff has been measured 16 confirms that BPA transfer does occur but the overall consumer exposure is low. This work is further evidenced by Oldring et al who worked on a model to estimate BPA-diglycidyl ether exposure, concluding that exposure for UK consumers was well below the tolerable daily intake for all derivatives. 17 With BPA non-intent (BPANI), migration studies have been carried out on coatings of similar chemistry, 18,19 as is common with all new coating formulations, with a multitude of analytical techniques used such as gas chromatography and mass spectrometry. 20 While these methods of interaction are relatively well reported, the effect of the simulant solution/foodstuff on the adhesion of a coating to the substrate, or the substrate itself post-retort, has not been reported in depth. Some work has been conducted exploring the delamination mechanism of an epoxy coating from a steel substrate, using electrochemical impedance spectroscopy (EIS) and X-ray photoelectron spectroscopy (XPS) to determine than an acid treatment improves the resistance against cathodic delamination, 21 most notably in a NaCl solution. The purpose of this study is to investigate the effects of the sterilisation (retort) process on both the substrate and the organic coating, attempting to understand the interaction between the two at the interface, 8 or indeed where any failure may occur.

| MATERIALS AND METHODS
Substrates used were ECCS, supplied by Tata Steel, and a developmental chromium-coated steel substrate (294269M), also supplied by Tata Steel. The developmental substrate was electroplated using Cr (III) species in the electroplating bath, rather than the Cr (VI) species used in ECCS production. 294269M was produced on a full-scale industrial electroplating line during parameter optimisation tests and is hence a developmental only substrate, not commercially available.
In each case, the bath chemistry and operation parameters of the electroplating process were manipulated to provide a controlled combination of surface species. Neither substrate has free Cr (III) or Cr (VI) present on the surface after electroplating, and chromium is then present in the form of Cr metal, oxide, or hydroxide; the details of which are shown in Table 1. 294269M shows increased levels of chromium oxide and lower values of metallic chromium when compared with ECCS, while also showing some evidence of iron at the surface.
Coatings used were an epoxy-phenolic (EP) lacquer and a polyester BPA non-intent (BPANI) lacquer, both supplied by Metlac. The former representing incumbent lacquer chemistry, the latter representing commercial state of the art. For the purposes of this research, sheets of both substrates were coated with 10 to 12 μm of either the EP or BPANI coating, providing four substrate/coating systems. For testing, sheets were cut into samples 120 mm × 45 mm.
To simulate the sterilisation process, coated substrates were placed in sealed jars filled to 80% of total capacity with simulant solution ( Table 2). Simulant solutions were chosen as they represent common packing media for canned foodstuff; the associated foodstuff is  Time of flight secondary ion mass spectrometry (ToF-SIMS) analysis was conducted on a Scientific Analysis Instruments (SAI) MiniSIMS-ToF, using a gallium source to obtain positive secondary ion spectra in the form of images.
In total, each result, for each experiment conducted, represents an average of five samples providing a statistically valid measurement of each substrate/lacquer combination. Where the lacquer properties were to be examined independently of the substrate, the lacquer was coated to glass microscope slides which were unaffected by the retorting process. Further details of the experimental equipment and procedures are available in previous work. 11 In order to examine the impact of the simulant and the retort process, the lacquers were applied to glass slides and processed in the same manner as the steel substrates. In this way, it was possible to readily obtain a free standing film as the film could readily be detached from the glass, allowing the lacquer chemistry to be investigated independently from the steel substrate.

| XPS analysis
The changes identified in the findings of the ToF-SIMS are corroborated and by examining the XPS surface analysis for 294269M (Table 3), which also identifies changes in key surface species. Under DI water conditions, the chromium metal exhibits some hydration, while the 1% NaCl produces a significant increase in the iron oxide and metallic iron observed on the surface.
The other notable feature of Table 3 is the increase of Cr 2 O 3 for the citric and lactic acid for 294269M and the elimination of iron oxide from the surface with citric acid. In order to examine this mechanism, ECCS and 294269M substrates were subjected to retort in increasing concentrations of citric acid in a 1% NaCl solution. In contrast to increasing NaCl concentrations, there is an increase in the measured adhesion when the citric acid concentration is raised ( Figure 6). The increase is significant and raises the adhesion between the pre-retort dry adhesion. The role of citric acid as a surface passive has been reported in stainless steel processing 25,26 where it provides a more environmentally friendly and safer passivation treatment compared with conventional nitric acid passivation. The passivation treatment results in preferential dissolution of iron oxide 25,27 and oxidation of  rosion is clearly illustrated in Figure 3 and Table 3 where the exposed iron levels are greater than the other simulants. For

| Effect of oxygen on adhesion
Corrosion of chromium-coated steel substrates is enhanced by the presence of oxygen, which acts as an oxygen source in the simulant. 29 Thus, by elimination of the oxygen from the simulant fluid, it was possible to examine the role of oxygen in the corrosion of the substrate.   therefore plays an appreciable role in the failure mechanism and thus reinforces the ToF-SIMS and XPS analysis in Figure 3 and Table 3. This result has an impact for can fillers as dissolved oxygen in the foodstuff will tend to increase the corrosion and hence substrate/lacquer adhesion. Similarly, any free air head space in the can will have an impact on the substrate/lacquer adhesion. From a material testing viewpoint, any DI water will need to be left for a period of time prior to use in a retort environment in order to equilibrate the dissolved O 2 in the DI water. Given the liquid volume to substrate surface area employed in the testing, any corrosion will not be limited by the quantity of dissolved oxygen in the liquid. In a sealed can where the liquid volume to surface area is far lower, the availability of dissolved oxygen may become a limiting factor in corrosion. 30,31 In this instance, the failure force measured is likely to be overestimated in this research. In the environment of a can, the surface area of coating and substrate is much higher than that of the testing regime, where a small number of samples are tested in a glass container, limiting the substrate that is available for oxidation.  This is attributed to the presence of the chloride ions causing corrosion at points in the surface where iron is present at the interface. 31 The thermal expansion of the gaseous corrosion product then leads to organic/metal separation. At a lower pH, according to the revised pourbaix diagrams at elevated temperatures 32 for Cr 3+ , Cr 2 O 3 is no longer the stable state of Cr 3+ . Instead, chromium is more likely to be in the aqueous Cr 3+ state or in the CrOH 2+ state. This suggests that the mechanisms of corrosion are different when an acid is added to the system; at low pH, the chromium oxide layer is likely to be affected, whereas at neutral pH, the oxide layer is stable, and hence the corrosion mechanism is likely to be related to the exposed iron in this developmental substrate. In the case of 1%NaCl/1% acetic acid, a corrosion effect is observed with no micro delamination shown in SEM; the same can be said of the samples exposed to 1% lactic acid.

| Citric acid effect
The effect of citric acid, demonstrated in Figure 7, is more pronounced on 294269M than the ECCS, and this attributed to the higher chromium level which is applied which allows greater conversion to chromium oxide and thus an increased number of lacquer bonding sites. 33 This mechanism also results in higher substrate/lacquer adhesion on 294269M compared with ECCS which is in contrast to that   As reported previously, to provide a statistically valid and reproducible quantitative measurement, scratch testing has been used which by nature of the testing will take into account both adhesive and cohesive forces.

| FTIR analysis
The chemistry that is observed by FTIR shows (Figure 10

| DISCUSSION
A number of failure mechanisms of novel materials, designed to be compliant with future legislation, have been identified. Their performance is generally lower than the incumbent technology and the failure mechanism changes with the simulant (or the food being packaged). The interaction which occurs at the interface between the simulant and surface chemistry determines the failure mechanisms, and this is summarised in Table 4. The work has shown the novel substrate performs less well that the incumbent technology. This is primarily associated with the

| CONCLUSIONS
The investigation has demonstrated that there exists a significant interaction between the chemical nature of the canned foodstuff, the packaging materials, and the lacquer/substrate lacquer adhesion.
While past substrate/lacquer combinations have been robust, those  give differing results but are out of the scope of this work.