The Kinetics and Mechanism of Atmospheric Corrosion Occurring on Tin and Iron-Tin Intermetallic Coated Steels II. Filiform Corrosion

This item is brought to you by Swansea University. Any person downloading material is agreeing to abide by the terms of the repository licence. Copies of full text items may be used or reproduced in any format or medium, without prior permission for personal research or study, educational or non-commercial purposes only. The copyright for any work remains with the original author unless otherwise specified. The full-text must not be sold in any format or medium without the formal permission of the copyright holder.

The increasing price of tin in the commodities market has resulted in a need to reduce the amount of tin used in tinplate materials (tin coating weight) for the corrosion protection of packaging steel whilst retaining adequate corrosion resistance. 1 One approach to doing this has been to diffusion anneal the unflowed tin (<1g.m −2 ) in a reducing atmosphere at temperatures above 500 • C 2,3 in such a way that almost all (preferably 90-98%) 4 of the free tin is converted to an iron-tin (FeSn) intermetallic as discussed during Part 1 of this paper. 1he aim of the current paper is to present a detailed study of the role of free tin and iron-tin intermetallic layers in resisting atmopheric corrosion, specifically filiform corrosion (FFC), as it affects the external surface of lacquer coated packaging material.
FFC, which produces 'threadlike' corrosion product deposits, tends to propagate from breaks or penetrative defects in organically coated metal products.It is widely accepted that filament advance involves anodic undercutting at the 'active head.'This process is driven by differential aeration arising from facile O 2 diffusion in the filament tail, which consists of dry corrosion products as shown by schematic a of Figure 1 [5][6][7][8][9][10][11] .Elsewhere it has been suggested that delamination of the protective layer from the substrate is due to cathodic mechanisms in the vicinity of the head. 12Senöz et al. used SKPFM to show that the anodic reaction within the head drives cathodic delamination of the coating at intermetallic particles within aluminum alloys.It was subsequently concluded that the particles within the vicinity of the head determined the direction of FFC 13,14 Williams and McMurray observed an area of cathodic delamination proceeding the head but contest its function as a primary cathode, claiming it plays no role in the mechanism of FFC advance. 15FC has previously been observed on exposed steel on score lines of packaging material lids, causing perforations from the outside in. 16,17Morita and Yoshida studied the FFC behavior of lightly tincoated steel for welded cans in relation to free tin, where free tin coating weight was defined as total tin minus alloyed tin. 18However, their results were complicated by the presence of a chromium based coating.
Here the time-dependent extent of FFC on a range of technologically important tin and iron-tin intermetallic coated packaging steels has been investigated optically and electrochemically.An explanation of the findings is given, firstly, in terms of the galvanic polarity of the corrosion cell formed between relevant phases and secondly in terms of the relative susceptibility of the various phases to anodic dissolution.

Experimental
Materials.-Ironfoil of 0.15 mm thickness and 99.5% purity and tin foil of 0.25 mm thickness and 99.8% purity were obtained from Goodfellow Cambridge Ltd.Mild steel with three different types of tin based coating were obtained from Tata Steel Packaging.The first coating consisted of unflowed porous pure tin of coating weight 2.8 g.m −2 .The second coating consisted of reflowed FeSn present at coating weights 0.44 g.m −2 and 0.88 g.m −2 .The third coating was again reflowed and consisted of FeSn 2 .The FeSn 2 coating was found to contain a small amount of surface free tin which was removed electrochemically by applying a controlled anodic current density in a 1 M HCl electrolyte 19,20 .The final FeSn 2 coating weights were calculated as 0.37 g.m −2 and 1.6 g.m −2 using chronocoulometry.
Polyvinyl butyral (PVB) and all other chemicals were obtained from Aldrich Chemical Co. and of analytical grade purity.All samples were cleaned and degreased using ethanol and distilled water before experimentation.
Methods.-In the case of electrochemical characterization experiments, coupons of approximately 40 mm x 30 mm were cut from large sheets to obtain a suitably sized sample.The sample was masked using extruded PTFE tape (type 5490 HD supplied by 3 M) which exposed a 10 mm × 10 mm area in the centre.Electrochemical measurements were taken using a Solartron 1280 Electrochemical Measurement Unit.A saturated calomel electrode (SCE) reference electrode was used to provide a fixed potential throughout the experiment.Open circuit potential (OCP) values were taken in a 0.1 M HCl electrolyte at 20 • C. The electrolyte pH was typical of that found within a FFC head. 6,7,21Potentiodynamic scans were conducted in a 0.6 mol.dm −3 NaCl electrolyte at 20 • C.Although the electrolyte pH was not representative of that within the FFC head it was considered that an acidic electrolyte would destroy samples.A platinum gauze counter electrode and a scan rate of 0.1667 mV.sec −1 were used.
The initiation and propagation of FFC was investigated on five different materials as shown in Figure 1b and followed a methodology described elsewhere. 15In the study of FFC on coated samples (Cell 3 -Cell 5) two types of sample were prepared.In the case of the first type the metallic coating was continuous.In the second type the metallic (tin or iron-tin intermetallic) coating was removed from half the sample surface.All samples were then solvent coated with 15% w/w ethanolic solution of polyvinyl butyral (PVB), molecular weight 70,000-100,000, lacquer using insulating tape height guides to give an air-dried thickness of 30 μm.Two types of experiments were carried out.In the first type FFC was initiated on the coating.In the second type FFC was initiated on the exposed steel substrate and allowed to propagate over the coated portion of the sample.In all cases a 10 mm line penetrative PVB coating defect was created by scribing the sample with a scalpel blade.FFC was initiated by introducing 2 μL of 2.5×10 −3 M aqueous FeCl 2 evenly over the length of the scribe using a glass microcapillary.After allowing the FeCl 2 to react with the exposed metal, and excess water to evaporate in air, samples were placed in an environmental chamber.The temperature was constant at 20 • C and a relative humidity of 93% RH was maintained throughout the experiment by allowing the atmosphere to remain in equilibrium with a reference solution comprising saturated aqueous Na 2 SO 4 10H 2 O.Samples were removed from the humidity chamber at intervals in time to carry out photography and computerized image analysis.The image analysis software (Sigma Scan Pro) was calibrated by specifying a pre-measured distance between two points and inputting real distance.The surface area of the FFC attack was measured as that occupied by corrosive discoloration.A value for a designated surface area was then given and the rate of propagation calculated.Six repeat experiments were conducted for each material and confidence limits (errors) correspond to ± one unit of standard deviation on the mean rate value.Optical micrographs of FFC were obtained using a Keyence VHX-700F digital microscope.Focused ion beam (FIB) milling was used to investigate a cross section of the FeSn coated sample to determine whether removal of the coating occurred during FFC propagation.The instrument used was a FEI Strata FIB 200 × P. A shallow trench was cut using a gallium ion beam and a standard cross section pattern ∼25 μm × 15 μm and ∼ 10-15 μm deep.The trench face was cleaned using a beam of lower beam current.The sample was tilted by approximately 45 • for imaging.   2 shows that anodic currents measured on FeSn are over ten times smaller than those observed for pure iron at the potential value ∼ 0 V vs. SHE.FeSn therefore has a significantly higher anodic overpotential near open circuit potential values.Pure tin achieves passivity fairly rapidly over this range of potentials; breakdown occurs at values of ∼−0.25 V vs. SHE.In comparison the overpotential to achieve passivity on pure iron or FeSn is much higher.The three zero current measurements observed in the case of iron indicates an active/passive transition suggesting the material is unstable, this being expected at the relevant pH. 22iliform results.-Pureiron cell (Cell 1).-FFC was observed on pure iron when initiated using a 0.0025 M FeCl 2 electrolyte.This finding is in agreement with the literature, the term 'filiform corrosion' first being used by Sharman in 1944 when studying lacquered steel surfaces. 7,15,21The catalytic properties of iron with respect to the oxygen reduction reaction (ORR) have been demonstrated during Part 1  of this paper and previously. 1,23,24The material is capable of maintaining an acidic electrolyte, the pKa for Fe (III) being 2.2, and is active at the relevant pH values. 22ure tin cell (Cell 2).-FFC was not observed to occur on pure tin when initiated using a 0.0025 M FeCl 2 electrolyte.The pKa for Sn (II) is 3.4 and Sn (IV) 2.2. 25 In theory, the head electrolyte is therefore acidic enough for FFC to be maintained and it is thus suggested that there is alternative reason that FFC is not observed.Tin is also active at the pH relating to that within the head electrolyte. 22With respect to the cathodic reaction, it has been stated that tin is associated with a high oxygen overpotential elsewhere. 26This is confirmed by results given during Part 1 of this paper which confirms a high overpotential of pure tin with respect to the ORR, when compared to pure iron and the FeSn and FeSn 2 intermetallic coatings. 1in-Iron cell (Cell 3).-The visual appearance of tin coated steel of both experimental types, 6 weeks after initiation, are shown in Figure 3.

Results and Discussion
As can be seen, FFC was not initiated on the tin coating.This supports the work of Morita and Yoshida who found that FFC was retarded by high cathodic polarization with increasing levels of free tin. 18The Keyence VHX-700F digital microscope was used to acquire images of the boundary between the steel and intact tinplate region as shown in Figure 4. FFC was found to propagate <500 μm when initiated on the steel substrate.
OCP characterization shows that pure tin is not sacrificial to iron at pH 1, this being typical of that found within the FFC head. 6,7,218][29] However, in the present case the galvanic polarity of the iron-tin cell ensures the FFC does not propagate farther into the tinplate and remains on the steel substrate FeSn-iron cell (Cell 4) and FeSn 2 -iron cell (Cell 5).-FFC corrosion has previously been found to accelerate on FeSn 2 in the presence of free tin due to galvanic coupling and the formation of microcells. 18hese results were however made more complex by the presence of an additional chromium/chromium oxide coating. 18In the present case FFC has been shown to occur in the absence of both free tin and chromium.Figure 5 and Figure 6 show that FFC was both initiated on 0.44 g.m −2 FeSn and propagated onto 0.44 g.m −2 FeSn after ini- tiation on the steel substrate using 0.0025 M FeCl 2 .Figure 7 shows a comparison between FFC on 0.44 g.m −2 and 0.88 g.m −2 FeSn, 6 weeks after initiation.In the case of the former sample substantially more corroded area exists.Figure 8 shows images of FFC initiated on 0.37 g.m −2 FeSn 2 .
The increase in corroded area with decrease in coating weight is supported by Figures 9a and 9b, which show the linear growth rate of the sample area over which coating delamination had occurred as determined by image analysis of pure iron and FeSn and FeSn 2 intermetallic coatings over the experimental time periods.The linear growth rate was found to be (0.0353 ± 0.0032) mm 2 .hr−1 on 0.44 g.m −2 FeSn, and (0.0122 ± 0.0011) mm 2 .hr−1 on 0.88 g.m −2 FeSn, in comparison to (0.0551 ± 0.0050) mm 2 .hr−1 on iron.The linear growth rate was found to be (0.0296 ± 0.0027) mm 2 .hr−1 on 0.37 g.m −2 FeSn 2 , and (0.012 ± 0.0011) mm 2 .hr−1 on 1.6 g.m −2 FeSn 2 .The confidence limits (errors) shown correspond to ± one unit of standard deviation on the mean based on six measurements.For both intermetallics, the rate of FFC decreases with increasing intermetallic coating weight.It is therefore suggested that the coating acts as a barrier and in order for the FFC to progress forward the coating must be dissolved, thus exposing the iron substrate which, as shown  previously, is a strong electrocatalyst for the ORR and thus becomes the cathode. 1,23,24In the case of higher coating weights, increased currents are needed for longer periods of times in order for FFC to propagate, thus leading to a reduced rate.This theory is consistent with images obtained using FIB, during which the sample was initially milled after applying a platinum layer which acted as a protective sacrificial barrier.Figure 10a shows the intact coating, and Figure 10b a cross section within the tail region.As can be seen the anodic attack of the coating does not stop on the intermetallic, but continues into the substrate.
The E corr of FeSn and FeSn 2 was found to be 0.024 V and 0.028 V higher than that of pure iron, respectively.The small difference in values for both couples renders FFC thermodynamically feasible.It is therefore considered that the reduced propagation rate of FFC, when compared to iron, is due to the low electrocatalytic activity of FeSn with respect to anodic dissolution as shown in Figure 2. Whereas tin achieves passivity fairly rapidly causing a reduction in corrosion rate, no passivity is seen on pure iron or FeSn over the range of potentials investigated and thus their rate of anodic dissolution is higher.
The rate of FFC propagation was found to be identical, 0.012 mm 2 .hr−1 , on both 0.88 g.m −2 FeSn and 1.6 g.m −2 FeSn 2 .It is thus suggested that a higher coating weight is needed in the case of FeSn 2 to resist FFC to the same extent as FeSn.It is widely reported that there is no obvious mass transport limitation on FFC propagation and kinetics are suggested to be surface controlled. 6,7,11It is therefore proposed that the greater true surface area of FeSn 2 , as described during Part 1 of this paper, allows greater kinetic currents and thus FFC propagation rates are increased 1 .

Conclusions
A systematic optical and electrochemical study has been completed to show that; r Filiform corrosion (FFC) could not be initiated on pure tin.r FFC was not observed to initiate on steel coated with unflowed pure tin.r FFC was observed on both FeSn and FeSn 2 , propagation rates being lower than that on pure iron and decreasing with increasing coating weight in both cases.
r Corrosion of both intermetallic and underlying substrate occurs during the FFC of FeSn.
An electrochemical study has been completed to show that; r Pure tin was not sacrificial to pure iron at the relevant pH 1. r FeSn and FeSn 2 intermetallics were not sacrificial to pure iron at pH 1.

It is proposed that;
r In all cases wholly or partially exposed iron substrate acts ca- thodically.
r The galvanic polarity of the iron-tin cell created did not permit the propagation of FFC at the corresponding pH.r The reduction in FFC propagation rate on FeSn and FeSn 2 , when compared to iron, was due to the low activity of tin and iron-tin intermetallics for anodic dissolution.

Figure 1 .
Figure 1.a.) Schematic showing filiform corrosion mechanism.b.) Schematics showing the five galvanic couples on which the initiation and propagation of filiform corrosion was investigated.
Electrochemical characterization.-OCPcharacterization Results.-Allmaterials were characterized in terms of their OCP at 20 • C in a 0.1 M HCl electrolyte, the results being shown in Table I.The confidence limits (errors) shown correspond to ± one unit of standard deviation on the mean, on the basis of three repeat measurements.Potentiodynamic results.-Potentiodynamicscans were conducted in a 0.6 mol.dm −3 NaCl electrolyte at 20 • C. A scan rate of 0.1667 mV.sec −1 was used and results are shown in Figure 2. Figure

Figure 3 .
Figure 3. Photographs of samples taken after 6 weeks showing that a.) FFC could not be initiated on tin coated steel and b.) FFC did not propagate into tin coated steel when initiated on a steel substrate, using 0.0025 M FeCl 2 .

Figure 4 .
Figure 4.An optical microscope image showing the propagation of FFC onto the intact tinplate region from the steel substrate on which it was initiated using 0.0025 M FeCl 2 .

r
During the FFC of the FeSn and FeSn 2 intermetallic coated steel the coatings became the site of anodic metal dissolution.

Figure 10 .
Figure 10.SEM image showing a.) intact FeSn coating and b.) substrate corrosion in a trench cut from a tail of FFC initiated on 0.88 g.m −2 FeSn.