Mineralogical Transformation and Electrochemical Nature of Magnesium-Rich Primers during Natural Weathering

Magnesium-rich primers (MgRP) have generated great interest as a promising alternative to chromium-based primers for the protection of aluminum substrates but their performance during exterior exposure has not been well documented. This paper focuses on the evaluation of MgRP during natural weathering to gain insight into its mineralogical phase transformation and electrochemical nature. Control studies were conducted on Mg and AA2024-T3 coupons. The results indicate that Mg particles in MgRP transform into a variety of hydroxide, carbonate, and hydroxy carbonates. During natural weathering, CO2 inhibited the dissolution of both Mg and AA2024-T3 as a result of protective carbonate layer formation in the coating.


Introduction
Corrosion protection of aluminum alloys by Mg-rich primer (MgRP) is based on the principle of sacrificial cathodic polarization of aluminum by Mg.Mg (E Mg+2 /E Mg = −1.6V vs. SCE) is more active (i.e., anodic) in the galvanic series than aluminum (E Al+3 /E Al = −0.6V vs. SCE) and AA2024-T3 alloying constituents.Mg can serve as a sacrificial anode by depressing the natural corrosion potential of aluminum to a value where it does not favor corrosion.MgRPs are considered an environmentally OPEN ACCESS responsible alternative to chromium-based primers for protecting aluminum structures.MgRPs contain sufficient Mg particles to maintain electrical contact with the aluminum substrate as well as each other.It is important to ensure that surface treatment, if any, on the aluminum does not create an electrical insulation between the Mg-rich primer and the aluminum substrate.Options to reduce the electrochemical activity of Mg in MgRP include the use of Mg alloys (with Al and Zn) [1] or forming a layer of Mg oxide/hydroxide/carbonate [2] at the expense of some reduction in the cathodic properties of pure Mg.Both these approaches decrease the galvanic protection offered by pure Mg.
MgRP stability is dependent on the nature of Mg particles (pure Mg, Mg alloy [1], Mg with oxide/hydroxide/carbonate layer [2] on it), coating matrix, insulating or non-insulating [3], and the environment to which it is exposed [4].Panels coated with MgRPs have performed very well during outdoor exposure evaluation on test coupons at various sites across the US.However, concerns have been raised about premature blistering of MgRPs in aqueous and marine environments.It has also been observed that MgRPs fail rapidly and exhibit heavy blistering very early on in salt spray tests (ASTM B117), which is still a key test in certifying coatings for corrosion protection.In our previous work [4], we investigated the behavioral dichotomy of MgRP exposed to salt spray testing (ASTM B117) and natural weathering (in a non-marine, non-urban environment).The study revealed the presence of a thin and porous Mg hydroxide layer in primers exposed to salt spray, and a thicker, protective hydroxyl Mg carbonate (nesquehonite, MgCO 3 •3H 2 O) layer in primers exposed to natural weathering.MgRPs exposed to dry atmospheric conditions exhibit excellent blistering resistance but marine conditions are inherently not conducive to Mg carbonate formation at a rate rapid enough to protect the primer against blistering.
Upon extended interaction with Mg, atmospheric concentration of CO 2 (~420 ppm) results in a protective layer of Mg carbonate or hydroxy carbonate.The effects of CO 2 interaction with Mg are dependent on CO 2 concentration, interaction time, and electrolyte nature (presence or absence of chloride).Lin et al. studied the role of CO 2 in the atmospheric corrosion of AZ91 Mg alloy in the presence of NaCl and concluded that CO 2 inhibited NaCl-induced corrosion by generating slightly soluble hydroxy carbonates that provided a partly protective layer on the surface of the Mg alloy [5].Cathodic polarization studies on Mg-rich primer coated AA2024-T3 alloy [6] indicated that CO 2 in high concentrations buffers the pH on the AA2024-T3 surface and diminishes the extent of basic corrosion.It was also determined that the atmospheric concentration of CO 2 is not sufficient to prevent cathodic or basic corrosion.Mg particle dissolution results in basicity at the interface and is a primary requisite to cathodically protect aluminum from corrosion.Lindstrom et al. [7] reported that sodium chloride-induced atmospheric corrosion of Mg alloys (AM20, AM50, and AZ91) was inhibited by ambient concentrations of CO 2 .Corrosion rates were about three times greater in the absence of CO 2 than in the presence of 350 ppm CO 2 .The inhibitive effect of CO 2 was partly attributed to the formation of a slightly protective carbonate-containing film on the alloy surface.Blücher et al. [8] reported that the NaCl-induced atmospheric corrosion of aluminum alloy AA 1070 is 10-20 times faster in CO 2 -free humid air compared to air containing ambient levels of CO 2 .The corrosion inhibition by CO 2 was explained on the basis of the neutralization of surface electrolytes.In a subsequent paper, Blücher et al. attributed the passivating effect of CO 2 on the NaCl-induced corrosion of AZ91D to the Mg 5 (CO 3 ) 4 (OH) 2 •5H 2 O layer formed on the surface of AZ91D [9].
The inhibitive effect of CO 2 on Mg alloy AZ 31B was investigated via potentiodynamic polarization and electrochemical impedance measurements in NaCl solutions of different concentrations that were saturated with CO 2 [10].The corrosion current decreased and the corrosion potential shifted to a positive direction with increasing immersion time.Passive behavior was detected after eight hours of immersion and the passive range increased with increasing immersion times.The surface products on AZ91 were identified as MgO, Mg(OH) 2 and Mg 2 (OH) 3 Cl in NaCl solution, MgO, Mg(OH) 2 and Mg 2 (OH) 2 CO 3 in water saturated with CO 2 (without NaCl), and MgO, Mg(OH) 2 , Mg 2 (OH) 2 CO 3 and Mg 2 (OH) 3 Cl in NaCl solution saturated with CO 2 .The corrosion rate increased with increasing NaCl concentration in the presence and absence of CO 2 .The corrosion rate in NaCl solution saturated with CO 2 was found to be higher than that in NaCl solution purged with inert gas.
In this investigation, we sought to determine the mineralogical transformation occurring on Mg particles (in Mg-rich primer) during natural weathering, the cathodic nature of Mg-rich primer (treated vs. untreated), and the impact of atmospheric CO 2 on AA2024-T3 and the electrochemical nature of Mg.

Behavior of MgRP (MgPUnt) in Laboratory Environment
Figure 1a reveals the dichotomous behavior of Mg-rich (MgPUnt) primer (on AA2024-T3 substrate) in an accelerated corrosion test environment (ASTM B117) and natural weathering (at Hattiesburg, a non-marine, non-industrial environment).Panels exposed directly and without carbonic acid pretreatment to the salt fog environment exhibited blister formation within four hours of exposure.Mineralogical studies using Scanning electron microscopy (SEM)/Energy dispersive X-ray spectroscopy (EDS) and Fourier transform infrared (FTIR) spectroscopy indicated the formation of brucite (Mg(OH) 2 , flower-like morphology) on panels placed in the salt fog and nesquehonite (MgCO 3 •3H 2 O, needle-like crystals) on panels exposed to natural weathering.
Mg-rich (MgPUnt) primer was applied on metal panels (AA2024 T3, cold rolled steel R-36, aluminum Al-36, and phosphated cold rolled steel) as well as polymeric sheets (Noryl ® GTX) and dried at ambient for seven days before being placed in the salt-spray chamber (ASTM B117) to distinguish the influence of the substrate on blistering.As expected, blister formation occurred rapidly on all substrates (Figure 1b).This supports our hypothesis that blister formation is due to the electrochemical instability (high electrochemical reactivity) of Mg powder in the highly humid and chloride ion-rich environment of the salt fog chamber.

Behavior of MgRP in Marine Exposure Site
Mg-rich (MgPUnt) primer coated AA2024-T3 panels were exposed at Daytona Beach, Florida (USA).Needle-like crystalline features, similar to nesquehonite crystals, were observed on the primed panels after seven days of ambient beachside exposure.This observation supports the fact that Mg particles in Mg-rich (MgPUnt) primer interact with atmospheric CO 2 to form Mg carbonate/hydroxy carbonate as seen in the laboratory specimens.Figure 2 is an overlay of the optical image and FTIR signature of typical needle-like crystals of nesquehonite on MgPUnt surface (Figure 2a), Mg cube (Figure 2b), and in treated Mg powder.Peak assignments are presented in Table 1.The FTIR signature (Figure 3) of the MgPUnt surface shows a sharp peak around 3690 cm −1 that is attributed to O-H stretching vibration of Mg(OH) 2 , while the peaks around 3644 and 3502 cm −1 are due to stretching vibration bands of the magnesite (OH)− and H 2 O, respectively.The band at 1643 cm −1 is identified as HCO 3 − stretching vibration, the sharp CO 3 −2 stretching vibration peaks around 1473 and 1411 cm −1 arise from nesquehonite and hydromagnesite, the broader band around 1650 cm −1 is attributed to HCO 3 − stretching vibration, the band 1083 cm −1 is assigned to Mg-O stretching vibration of MgO, and the peak seen at 849 cm −1 is due to the CO 3 −2 bending vibration band of nesquehonite and hydromagnesite.
The brucite peak (3690 cm −1 ) was only observed on the panel that had been exposed at Daytona Beach for seven days.After 30 days exposure, the panel exhibited a new carbonate peak at 864 cm −1 that confirmed the progressive conversion of Mg to brucite, nesquehonite, and hydromagnesite.Coated MgPUnt panels exposed at the rain forest and marine sites in Hawaii for six months exhibited a brucite peak at 3695 cm −1 .However, no peaks were observed for carbonate formation (Figure 3d).
Figure 4 shows the X-ray diffraction (XRD) patterns of (a) bare AA2024-T3 and MgPUnt coated on AA2024-T3 and exposed at (b) Hawaii (marine exposure) and (c) Daytona beach (7 days and 30 days).XRD patterns (Figure 4b  This observation supports our hypothesis that Mg particles interact with CO 2 even at atmospheric concentrations to form various carbonates and hydroxy carbonates.The extent and kinetics of Mg-CO 2 formation and interactions depends on the environmental severity, i.e., humidity, level of salinity, CO 2 solubility, and presence of topcoat over Mg primer.

Electrochemical Response of MgRP (MgPUnt and MgPT)
The open circuit potentials (OCPs) of Mg cube, bare AA2024 T3, MgPUnt and MgPT were monitored for an hour in 3.5 wt% NaCl solutions to comprehend the galvanic coupling between the Mg particles in the coating and the AA2024-T3 substrate.The results are presented as an OCP vs. time plot in Figure 5.The OCP values of MgPUnt and MgPT show a mixed potential (i.e., OCPMg cube <OCPMgPUnt and OCPMgPT <OCPAA2024-T3) and lie in between Mg and AA2024-T3.The OCP values of MgPUnt and MgPT stabilize at significantly greater negative potentials than that of AA2024-T3, indicating the cathodic nature of treated Mg-rich primers.The variations in OCP values of our MgRP are in good agreement with those reported in the literature on Mg-rich primers.Treated MgPT is always less cathodic than MgPUnt.The extent of Mg particle treatment is critical since extended treatment times will result in loss of the cathodic protection afforded by Mg.

Impact of CO 2 on Electrochemical Response of MgPUnt
The effect of carbon dioxide on MgPUnt coated AA2024-T3 was characterized via polarization studies in 3.5 wt% NaCl solution with and without CO 2 .The representative OCP and polarization curves for coated panels are shown in Figure 6.Panels immersed in NaCl solution saturated with CO 2 exhibited a novel OCP (Figure 6a) and lower corrosion current density (Figure 6b).The OCP of MgPUnt in NaCl solution saturated with CO 2 started at −100 mV and slowly dropped to stabilize around −900 mV.At ambient (atmospheric) levels of CO 2 , MgPUnt coated panels exhibited a different trend in OCP, starting at −700 mV and increasing to −1.45 mV.The sharp decrease in potential after 900 seconds and 2500 seconds in the presence of saturated and ambient CO 2 levels, respectively, is likely to have been caused by the NaCl solution reaching the primer-metal interface and establishing a galvanic circuit between the AA2024-T3 substrate and the primer.The OCP polarization plots show that the corrosion potential (Ecorr) of the panels in NaCl solution saturated with CO 2 moved to the positive direction by approximately 200 mV relative to the panel immersed in NaCl solution with ambient CO 2 .The corrosion current density of the representative panel immersed in NaCl solution and saturated with CO 2 is far smaller than that of the panel immersed in NaCl solution with ambient CO 2 .Both OCP and polarization experiments suggest that CO 2 shifts the coating towards a lower corrosion current density and corrosion potential and supports our other observations.

Impact of Pre-Immersion Time in NaCl-CO 2 Solution on Electrochemical Response of Mg
In air, Mg forms a gray layer of Mg oxide on its surface that converts to Mg hydroxide in the presence of moisture.Mg hydroxide is stable only at basic pH.On the other hand, CO 2 is acidic in nature.To better understand the role of CO 2 on the Mg-rich primer, the electrochemical response of Mg was determined in 3.5 wt% NaCl under the following exposure conditions: (a) 3.5 wt% NaCl at ambient level of CO 2 , no pre-immersion, (b) 3.5 wt% NaCl saturated with CO 2 , no pre-test immersion (before electrochemical testing), (c) 3.5 wt% NaCl saturated with CO 2 , 18 h pre-test immersion, (d) 3.5 wt% NaCl saturated with CO 2 , 6 days pre-test immersion, and (e) 3.5 wt% NaCl at ambient level of CO 2 , 24 h pre-test immersion.CO 2 was bubbled through an aqueous solution of 3.5 wt% NaCl for 45 min to yield a saturated solution (pH value ~4.1).As expected, immersion of Mg coupons in this solution resulted in the formation of a gray layer on its surface.The effect of CO 2 on the electrochemical response of Mg in 3.5 wt% NaCl solution with saturated and ambient levels of CO 2 is presented in Figure 7.The OCP of Mg coupons moved towards a more noble direction with increasing immersion time in the presence of both ambient and saturated levels of CO 2 .With increasing immersion time, the anodic branches shifted in the positive potential direction indicating that (a) the anodic dissolution was progressively retarded, and (b) the formation of a passive film on the Mg surface that reduces the corrosion current density has occurred.OCP and polarization curves under conditions A, B, C, D, and E (Figure 7a) indicate that the pre-immersed samples exhibit lower corrosion current density and noble corrosion potential compared to samples without pre-immersion.Polarization plots of immersion under conditions A and E when compared to those of B, C and D suggest that even ambient levels of CO 2 can create a passive layer during the pre-test immersion which reduces the corrosion current density and moves the OCP towards a positive direction.
These observations are in good agreement with the observations reported by Qua et al. [10] on the corrosion behavior of AZ31B Mg alloy in NaCl solutions saturated with CO 2 .The authors suggested that passivation of AZ31B in NaCl solutions saturated with CO 2 will occur if the immersion time is >8 h and the passive range increased with increasing immersion time.Longer immersion durations facilitate the formation of thick insoluble hydroxy Mg carbonates which restrict and/or limit electrolyte access to the cathodic Mg surface and improve the inhibitive properties.

Impact of CO 2 on Electrochemical Response of AA2024-T3
To understand the neutralizing effect of CO 2 on AA2024-T3 substrates, OCP and polarization experiments were performed on AA2024-T3 panels in 3.5 wt% NaCl solution with ambient and saturated levels of CO 2 (Figure 8).It was determined that CO 2 shifted the corrosion potential towards the cathodic direction as had been observed by Maier et al. [6] and reveals that the beneficial effect of CO 2 is not only limited to Mg but also extends to AA2024-T3 with minimal CO 2 exposure in advance of being introduced to a corrosive environment.However, the behavior of MgPUnt, Mg, or AA2024-T3 in a NaCl solution cannot be directly correlated to the system's behavior during atmospheric exposure.Compared to bulk NaCl solution, the surface electrolyte experiences insufficient convection in atmospheric conditions and the natural surface layer tends to remain where that layer is formed unless leached away by water.Moreover, since the diffusion paths for oxygen molecules are short, oxygen supply is usually not rate-limiting for the cathodic reaction.Blücher et al. [11] found that ambient level of CO 2 strongly inhibits the NaCl-induced atmospheric corrosion of Al in humid air by buffering the hydroxide formed at the cathodes and preventing the formation of aluminates.

Materials
Mg cubes (19 mm × 19 mm × 6 mm) and Mg powder (325 mesh, 99.8% pure) were purchased from Alfa Aesar.Eponol ® resin 53-BH-35 (Hexion Specialty Chemicals, Inc., Norco, CA, USA) is an ultra-high molecular weight bisphenol-A-based epoxy resin supplied as a 35% solution (by weight) in a blend of methyl ethyl ketone and propylene glycol methyl ether (75:25 by wt), and reported to have a specific gravity of 0.934 at 25 °C and 26.6% volume solids.CO 2 gas cylinders and sodium chloride were purchased from Nordan Smith (Hattiesburg, MS, USA) and Aldrich Chemicals (Milwaukee, WI, USA), respectively.All chemicals were used as received.Aerospace grade aluminum alloy AA2024-T3 was purchased from the Q-Panel Company (Westlake, OH, USA).

Treatment of Mg Powder in CO 2 -H 2 O
Mg particles were immersed for 30 min in an aqueous carbonic acid (saturated solution of CO 2 -H 2 O, pH 4.1 at ambient) solution to develop a layer of protective Mg carbonate [2].The treated powder was used to formulate MgRP with Eponol.Mg cubes and AA2024-T3 coupons were also treated similarly in carbonic acid.Bare AA2024-T3 and Mg coupons (180 cm × 180 cm × 4 mm) were abraded with emery paper from 100-600 grit, cleaned with distilled water and acetone, and dried at ambient before use.

Preparation and Application of Mg-Rich Primer
MgRPs (with treated and untreated Mg powder) formulated at a pigment volume concentration (PVC) of 45% were prepared by blending Mg powder with Eponol overnight in a ball mill.The primers were applied on AA2024-T3 panels and Noryl GTX (a blend of polyphenylene oxide and polystyrene) plaques at 6 mils wet film thickness using an automated drawdown bar.The coated panels were kept in a dry box at ambient conditions for seven days prior to testing.The final dry film thickness was 75 ± 5 µm regardless of substrate type.The coated Al and Mg specimens were designated as: MgPT and MgPUnt (Mg-rich primer formulated with treated and untreated Mg particles, respectively), and MgT and MgUnt (Mg coupon treated and untreated).

Exposure Conditions
Coated AA2024-T3 panels were evaluated before and after exposure at marine and heavy rain forest sites in Daytona Beach, Florida and Hawaii, respectively.The electrochemical behavior of the coated AA2024-T3, AA2024-T3 substrate, and Mg coupons was investigated in an aqueous solution of 3.5 wt.% NaCl solution.

Characterization
The products on the surface of test specimens were characterized via Fourier transform infrared (FTIR) spectroscopy using a Digilab ® FTS 2000 (Digilab, Inc., Marlborough, MA, USA).Spectra were collected in transmission mode over a frequency range of 600-4000 cm −1 via 256 scans at a resolution of 8 cm −1 .XRD is a non-destructive analytical technique that reveals information about the crystallographic structure, chemical composition, and physical properties of materials and thin films.XRD studies were conducted on a Rigaku Ultima3 X-ray diffractometer (Rigaku Americas Corporation, The Woodlands, TX, USA) at a scan speed of 0.5°/min and scan angles between 10° ≤ 2θ ≤ 90°.
Scanning electron microscopy (SEM) utilizes an electron beam for sample surface imaging at typical magnifications of 25-50,000×.Changes in the surface morphology of the primer, before and after exposure were examined on an FEI Quanta 200 (FEI, Hillsboro, OR, USA) equipped with a Thermo System 7 energy dispersive X-ray detector (Thermo Fisher Scientific, Inc., Waltham, MA, USA).Optical images of samples were collected using Keyence ® VHX-600 digital microscopy system (Keyence Corporation of America, Itasca, IL, USA).
Electrochemical responses of the samples were evaluated by measuring their OCP, Tafel plot, and impedance values with potentiostat/galvanostat (VersaStatTM IV, Princeton Applied Research,

Figure 3 .
Figure 3. FTIR spectra of MgPUnt coated AA2024-T3 at ambient (a), after exposure at Daytona beach for 7 days (b) and 30 days (c); and at the rain forest site and marine exposure in Hawaii (d).
) of MgPUnt exposed at Hawaii confirm the presence of Mg(OH) 2 (brucite) and hydroxy Mg carbonate (nesquehonite).The panels exposed at Daytona exhibited the presence of brucite and nesquehonite.All the peaks in the XRD pattern of the MgPUnt are in good agreement with the peaks characteristic of hydroxy Mg carbonate (JCPDS 70-1177), Mg hydroxide (JCPDS 07-0239), and nesquehonite (JCPDS 70-1433).

Figure 5 .
Figure 5. Open circuit potential (OCP) values of Mg cube, bare AA2024 T3, MgPUnt and MgPT monitored for an hour in a 3.5 wt% NaCl solution.

Figure 6 .
Figure 6.Potentiodynamic polarization curves (a) and OCP (b) of MgPUnt coated AA2024 panels in 3.5 wt% NaCl solution saturated with CO 2 and ambient level of CO 2 .
− stretching of hydromagnesite 1418 cm −1 O-H stretching of H 2 O attached to hydrated Mg carbonate 3c 3644 cm −1 (OH) − stretching of hydromagnesite 3502 cm −1 O-H stretching of H 2 O attached to hydrated Mg carbonate 1473 and 1411cm −1 symmetric stretching mode of carbonate and 846 cm −1 asymmetric stretching mode of carbonate from hydromagnesite