Corrosion of Fusion Welded Magnesium-Aluminum Alloy, AZ31: Effects of Microstructure and Composition in Isolated Weld Zones

The corrosion behavior of tungsten inert gas (TIG) welded magnesium alloy AZ31B was explored using a variety of approaches. The main goals were to elucidate the microstructural and compositional factors that control corrosion and provide insight for corrosion mitigation strategies. In the initial work, the corrosion resistance of the as-received commercially pure Mg rod, AZ31B-H24 wrought base plate and AZ31B-H24 TIG welds were investigated by DC electrochemical methods as well as a 24 h open circuit potential (OCP) measurement complimented by electrochemical impedance spectroscopy (EIS) in 0.6 M NaCl solution. This approach provided a reliable indication of corrosion rate corroborated using four independent methods (1) mass loss, (2) EIS, (3) hydrogen gas collection and (4) Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) chemical analysis of exposure solutions. Analysis and comparison of corrosion rates of the wrought base and the isolated weld zones using these four parallel measurements enabled an unparalleled and reliable estimation of weld zone corrosion rate and provided a clear analysis of weld corrosion as a function of heat affected zone (HAZ), base and liquated base metal weld zones.
Weld structure and composition were characterized with optical microscopy and SEM. Significant changes to the alloy microstructure and corrosion behavior were observed after welding. The differences in corrosion rate can be attributed to microstructure attributes in the weld fusion zone (FZ) and HAZ. These include constituent particles which grow by Ostwald ripening within the α-Mg matrix during processing, eutectic solidification structures as well as recrystallization to form various grain sizes seen throughout the various weld zones. This dissertation is focused on first identifying and isolating the microstructural variables within the weld and then assessing how these factors in isolation affect the corrosion rate. These include mainly grain size, intermetallic particle (IMP) formation and distribution, solute segregation and crystallographic orientation. Specifically, the FZ was characterized by a randomized texture, large grain size and Al-Zn rich solidification boundaries. The HAZs were characterized by a basal texture, large recrystallized grains and larger, more widely spaced IMPs (in comparison to the wrought base). Further mechanistic understanding of corrosion as a function of weld zone was investigated by attempting to isolation and examine the effect of specific microstructural details on the corrosion rate.
The galvanic interactions between zones were also investigated using multichannel multi-electrode arrays (MMAs) comprised of sectioned weld zones in conjunction with in-situ time lapse video, scanning electron microscope (SEM) surface analysis and mixed potential theory analysis. The controlling region in weld corrosion was determined from time lapse videos by examining the dominate location of the hydrogen evolution reaction (HER) and darkening within the various weld zones which is considered characteristic of cathodic activation in Mg alloys. The variation in the anodic dissolution rates between weld zones and subsequent cathodic activation caused polarity reversal between electrochemically connected zones. The galvanic coupling between electrochemically connected weld zones was investigated diagnostically in three different environments, specifically 0.6 M NaCl, 0.6 M NaCl saturated with Mg(OH)2 and tris(hydroxymethyl)aminomethane (TRIS). The resulting corrosion morphology in each of these environments were compared and contrasted.
The first microstructural characteristics identified were IMP distribution and grain size. Due to the different electrochemical potentials between the Mg matrix and any impurities or IMPs which may form during processing, micro-galvanic corrosion occurs. The IMPs function as active cathodic sites. Moreover, “activated” zones in the Mg matrix around the IMPs contribute to the corrosion behaviour. This was rigorously assessed through the use of furnace heat treatments, model alloys, electron microscopy and electrochemical methods. The relationship between particle size, particle spacing, and the resultant corrosion behaviour was explained. Additionally, model samples were developed to study IMP effects in isolation from other metallurgical effects (particularly grain size and residual stresses), with particles simulated by Al electrodes embedded in a Mg matrix. Arrays of model Mg-Al electrodes were constructed using high purity Al as a surrogate for Al-rich IMPs and flush mounted in commercial purity Mg. The area fraction, size and spacing of these electrodes each altered the corrosion rate and cathodic reaction kinetics over a 24 and 48 hour immersion period at the open circuit potential. The grain size was ruled out as a governing factor controlling the corrosion rate. However, small, close spaced Al-rich IMPs or synthesized electrodes raised the overall corrosion rates on Mg. The decrease in the corrosion rate with increasing particle spacing could help explain the decrease in the corrosion rate in the HAZ (in comparison to the baseplate) due to the larger particle spacings determined to exist in these weld zones.
Another metallurgical characteristic to evaluate was the effect of crystallographic orientation on corrosion rate due to the variation in crystallographic texture in the FZ versus the HAZ and baseplate. The electrochemical dissolution of Mg indicates strong crystallographic dependence based on film thickness in chloride-containing, environments. For non-chloride containing, neutral pH environments which dissolve air-formed MgO and support very little film growth, such as TRIS and Ethylenediaminetetraacetic (EDTA), there was initially limited crystallographic orientation dependence on the corrosion rate which correlated with surface energy. The origins of the corrosion rate trend were investigated utilizing EIS. EIS constant phase elements were exploited to determine oxide thicknesses. In 0.6 M NaCl, the corrosion rate varied with the film thickness. The fastest corrosion rate occurred on the basal plane and the slowest corrosion rate occurred on low index, prismatic and pyramidal planes. This variation in the corrosion kinetics with crystallographic orientation in 0.6 M NaCl was directly proportional to MgO film thickness. In particular, the variation in the MgO and Mg(OH)2 thickness for faces with various crystallographic orientations may strongly alter the corrosion kinetics. The fastest corrosion rate was observed on the basal plane which correlates to the faster corrosion rate observed in the basal oriented HAZ and base material of the TIG weld.
The variation in the corrosion rate due to solidification structures formed during melting and resolidification of the FZ was also identified and characterized. Due to the high temperatures seen during welding and the differing solubility of the Mg and its alloying elements, solute-rich solidification boundaries can form in the FZ. To evaluate the corrosion of these regions with solidification structures and high solute content, the corrosion morphology was evaluated after exposure in 0.6 M NaCl. The corrosion rates for several die-cast Mg-Al alloys (AM50, AM50 and AZ91) were determined utilizing the four parallel methods used previously as these alloys enabled systematic evaluation of the effects of increased Al content and solidification boundaries on corrosion rate, as these alloys were deemed to be structurally similar to the FZ. The variation in the cathodic kinetics for the die-cast alloys were determined over 3, 24 and 48 hour immersion periods, at the OCP, in three electrolytes, unbuffered 0.6 M NaCl, 0,1 M TRIS, and 0.6 M NaCl buffered with TRIS. It was observed that the corrosion rate decreased with increasing Al content in spite of an increase in the phase fraction of IMPs. Variation in the cathodic activation with time for each of these environments is observed with lower levels of cathodic activation occurring in 0.1 M TRIS. The lower corrosion rate observed on the alloys containing large solute segregation correlates to the lower corrosion rate observed in the resolidified FZ.
An additional study mapped the actively corroding sites in the weld and base material. The nature of the galvanic couples within the material (between weld zones and at localized IMP sites) was investigated using scanning kelvin prove force microscopy (SKPFM) and scanning electrochemical microscopy (SECM) to map the potential differences due to various microstructural features across the weld. Variations in the activity on the solidification boundaries in the FZ as well as the IMPs in the HAZ can be mapped using these in-situ techniques to better define how the microstructure alters the corrosion rate for a heterogeneous metal system.
The outcome of this research provided a fundamental understanding of the Mg weld microstructure property paradigm to (a) understand metallurgical factors which control corrosion, (b) provide insight regarding attributes to maximize or minimize in weld microstructures for a combination of intrinsic and galvanic corrosion resistance. Examples of this are provided in the future work section.