Metallurgical and Electrochemical Characterization of the Corrosion of a Mg-Al-Zn Alloy AZ31B-H24 Tungsten Inert Gas Weld: Galvanic Corrosion Between Weld Zones

Mg-Al alloys are of interest for aerospace and automotive industries.^1-2  In particular, the effect of fusion welding on the intrinsic corrosion properties of magnesium (Mg) is a concern for the application of Mg structures should this joining strategy become of interest. However, the metallurgy of these fusion welded structures must first be understood at the multiple length scales that affect corrosion properties.

The corrosion behavior of Mg and its alloys is complex as a result of the microgalvanic cells, which are formed in the material as a result of small differences in electrochemical potential between the α-Mg matrix and secondary phases.^3-5  The variation in this galvanic coupling is highly dependent on the alloying elements added into solution above or below their solubility limits⁴  and the intermetallic particles (IMPs)⁶  in the material. The addition of fusion welding techniques, such as tungsten inert gas (TIG) welding, can further complicate this issue by forming various weld zones in the material, each of which can have individual electrochemical potentials resulting from differences in grain size, crystallographic texture, solidification structures, and IMPs as described in previous work.⁷ 

In a fusion weld, different zones are formed during welding (Figure 1). The fusion zone (FZ) is completely melted and resolidified.⁸  This zone contains fine equiaxed grains, as well as secondary phases which consist of IMPs^9-10  and solidification structures, and is typically characterized by a randomized texture.¹¹  Also, several heat affected zones (HAZs) are formed. These are compositionally similar to the starting wrought base and retain the basal texture seen in AZ31 (UNS M11311)⁽¹⁾ wrought base¹¹  but have variations in grain size and particle distribution as a result of non-uniform heating and cooling temperature variations across the weld.

These small variations in the metallurgy can cause dramatic changes in the electrochemical potentials of the individual weld zones and lead to the development of microgalvanic cells which accelerate the corrosion rate of the overall weld. Therefore, when these regions are tested in electrical contact, the corrosion rate is expected to be higher than if the weld regions are tested in isolation.⁷  The resulting galvanic corrosion rate is studied herein.

Galvanic corrosion rates between the wrought base and the full weld zone have been studied,^12-13  but much of this literature focuses on comparing the corrosion rate of these weldments in salt fog or atmospheric tests and does not investigate the corrosion rate of these welds in full immersion testing using electrochemical and other diagnostics.¹²  While there is a significant quantity of literature on Mg,^14-21  most focus on the isolated weld zones using techniques such as microcapillary techniques,²²  solid state welding,^17-18,23-24  or small area test measurements.⁷  However, once coupled, a pertinent focus is the galvanic couple potential relationships and polarity, with respect to anode or cathode, of each zone over time with respect to each other.

In current work as well as the study herein, the phenomenon of anodically-induced cathodic activation has been cited to occur in the Mg system, playing a role in open-circuit corrosion and contributing to the explanation of the negative difference effect.^25-27  Several potential explanations have been proposed including transition metal enhancement²⁶  on the sample surface, formation of a semi-protective surface film,^27-28  and Al replating.²⁹  This enrichment²⁶  or replating²⁹  will increase the cathodic reaction rate over time and ennoble the sample surface. It has been proposed that increased anodic activity on the sample surface can occur with time as a result of enhanced cathodic processes.^30-31  The result of this with time is an increase in the hydrogen evolution reaction (HER) on the sample surface²⁷  at strong cathodes.³⁰  The increase in cathodic kinetics will also affect the interactions and galvanic relationships between weld zones and the resulting polarity of each zone. This important aspect has never been considered in weld zone corrosion.

Galvanic couples between electrochemically connected weld zones can cause polarity reversal resulting from rapid changes in surface chemistry on the respective anodes and cathodes on the exposed weld. For instance, elements may be replating²⁹  or enriched.³²  Many material systems undergo a polarity reversal during galvanic corrosion. In particular, the polarity reversals between steel and Zn^33-34  and steel and Al³⁵  have been studied. Polarity reversal is typically a result of a change in the surface condition of the coupled metals.³⁶  This polarity reversal is also highly dependent on the environmental conditions such as dissolved ions, pH, and time of immersion. In the steel-Zn couple, passivation of the zinc surface and a sufficient amount of dissolved oxygen leads to this polarity reversal.^16-17  In addition, in the steel-Al couple, polarity reversal occurs in environments where Al alloys are used as sacrificial anodes for steel. It is well known that the corrosion potential depends on the level of passivity on the alloy³⁷  and this passive film growth can alter the polarity between steel and Al when coupled.

For Mg, this alteration of the surface condition can occur as a result of the rapid formation and breakdown of the semi-protective and porous film.^38-41  Similarly, it has been hypothesized that Mg and its alloys suffer from enhanced cathodic activity as a result of the enrichment of transition metals.^26,32  In this work and previous work by the authors,⁷  cathodic activation has been shown to occur in the different weld zones, which causes initially anodic weld regions to change polarity and become cathodic weld regions with increased immersion time.

Multichannel multi-electrode arrays (MMAs) are ideal for on-line real-time monitoring of localized corrosion rate.^42-43  The MMA enables the measurement of the electrochemical current of each addressed electrode within the array, separately and simultaneously. The use of this technique enables investigation of a simplified geometric weld scenario in order to define the corrosion current through a specific spatial distribution. Work on MMAs in galvanic scenarios has been used previously^42,44  and has been successful at determining the corrosion rate related with localized phenomenon such as crevice corrosion.^42,45  In particular, MMAs have been used to observe the corrosion rate in systems which show non-uniform corrosion characteristics and in systems in which polarity reversal is observed when connected electrochemically, such as the polarity reversal seen in Zn-Steel galvanic couples.⁴⁶ 

The objective of the work herein is to understand the galvanically controlled corrosion in fusion welded AZ31B-H24. The already published work is related to understanding how to determine the corrosion rate of the wrought base⁴¹  form of these alloys and the isolated weld zones,⁷  but further understanding the galvanic coupling effects within the weld is also of interest. Both whole welds and weld array electrodes were investigated. In addition to the electrode configuration, various corrosion environments are investigated. Better understanding into the variables that have an impact on the galvanic corrosion in Mg welds will enable industrial application where this joining strategy is of interest.

AZ31B-H24 sheet was provided by Magnesium Elektron^†. All compositions are reported in wt%, with the actual compositions provided by QUANT^† (Quality Analysis and Testing Corporation) via inductively coupled plasma optical emission spectroscopy (ICP-OES) analysis, as given in Table 1.

All samples were fabricated by TIG welding a sandwich structure as shown in (Figure 1). The top plate was drilled and countersunk at 45° to approximately 0.635 cm in diameter and welded to the bottom plate along the periphery of the countersunk hole. The Miller Syncrowave 200^† TIG welder was operated at 125 A under argon gas at 48.3 kPa with a flow rate of 18.5 L/min, and air cooled. This process created large metallurgical weld zones for easy study both in isolation and in the MMA. All samples were prepared on material freshly ground through 240 grit to remove any oxides formed before processing. Irradiated regions were investigated and compared to the baseline starting materials. All samples were examined along the surface, polished through colloidal silica, and etched with a picric acid etch (3 g picric acid, 30 mL acetic acid, 100 mL ethanol, and 15 mL distilled water) to determine the grain size and microstructure.

Samples were analyzed with optical microscopy as well as scanning electron microscopy (SEM) using a FEI Quanta 650^† microscope. Compositional analysis was performed using energy dispersive spectroscopy (EDS) methods.⁴⁷  Images were recorded at a working distance of 10 mm while operating at an accelerating voltage of 5 kV. Electron backscatter diffraction (EBSD) was performed while operating at an accelerating voltage of 20 kV. The composition of the IMPs, solidification boundaries, FZ, HAZs, and wrought material were determined using rigorous EDS compositional analysis with ZAF corrections (where Z is the atomic number correction, A is the absorption correction, and F is the fluorescence correction) on the Aztec^† software tool.⁴⁷  The distribution and area fraction of the Al-Mn-Fe IMPs, as well as Al-Zn solidification boundaries, were determined using the ImageJ^† software package.⁴⁸  At least three measurements were analyzed for both grain size and composition, and an average was reported. Standard error from these runs has been reported where appropriate.

Samples were immersed for 3 h, 24 h, or 48 h at open-circuit potential (OCP) with the weld zones electrochemically connected in aerated 0.6 M NaCl before isolating each zone using electroplaters tape with a 1.23 mm² hole. This range in immersion time allowed for time-based development of the cathodic activation of each zone as a function of immersion time of the galvanically connected weld zones. The cathodic behavior of the weld zones was determined by a cathodic potentiodynamic polarization scan ranging from 50 mV above OCP to −2.3 V below OCP in a downward sweep at a rate of 1 mV/s. A Pt mesh was used as a counter electrode and a saturated calomel electrode (SCE) was used as the reference electrode.

The corrosion morphology was observed near the interface between the weld zones to determine how the corrosion morphology varied between weld zones when electrochemically connected in the solution. SEM micrographs of each of the weld regions were taken before and after immersion in 0.6 M NaCl, 0.6 M NaCl saturated with Mg(OH)₂, and tris(hydroxymethyl)aminomethane (TRIS) for 3 h at OCP. The starting pH of these environments was 5.3, 10.2, and 7.0, respectively. The pH after immersion for 24 h of both NaCl mixtures was approximately 11. The TRIS solution functions as a buffer and retained the same pH over the full immersion period. The conductivity of the 0.6 M NaCl is 7.6 × 10⁻² (Ω·cm)⁻¹, while the conductivity of 0.1 M TRIS is 1.97 × 10⁻⁴ (Ω·cm)⁻¹. Following testing, samples were cleaned following the ASTM Standard G1-03 using chromic acid solution (200 g/L CrO₃) to remove any corrosion product.⁴⁹  It should be noted that, after immersion in the chloride environments, SEM analysis (not included) showed a large amount of Mg(OH)₂ corrosion product similar to that observed in other works.^30,50-51  Raman spectroscopy and electrochemical impedance spectroscopy were used to detect surface films. Before the initial corrosion, the sample was marked using a Vickers hardness tester to enable examination of the corrosion morphology in precise locations. The progression of the corrosion morphology was studied with time using a DinoLite^† optical microscope with time-lapse video and SEM imaging. A software package was used to calculate the area-fraction of the sample surface that was corroded with time.⁴⁸  The area-fraction was determined at 3 h, 12 h, and 24 h of immersion using selected SEM images in the isolated case and time-lapsed optical videos for the galvanically connected case. The standard error is reported.

The galvanic currents of coupled weld zones were determined using a zero resistance ammeter (ZRA) in a flat cell.^52-53  The samples were coupled with one isolated weld zone as the working electrode (WE) and another isolated weld zone as the counter electrode (CE). The FZ was always assigned as the CE for each galvanic couple and the individual HAZ or base was always designated as the WE. Therefore, a positive current indicates that the HAZ was the anode. The cathode to anode ratio was kept at 1:1 with each of the weld zones isolated using electroplaters tape with a 1.23 mm² hole. All galvanic couples were tested at least three times and the reproducible trends are reported herein.

MMAs were also used to determine the current from the galvanically connected weld zones. The weld was sectioned into 12 sections where each electrode had a surface area of 3 mm² (Figure 2). The individual weld zones were mounted in epoxy resin and separated using a 0.13 mm shim to fully electrically isolate individual weld zone sections while still keeping the spatial resolution on the same scale as the actual weld. In a realistic weld, the weld zones have different sizes; therefore, this was taken into account on the design of the MMA by controlling the number of channels used. Large weld zones were assigned more channels, while smaller weld zones were assigned fewer channels according to their approximate size in the weld. The approximate area fractions of the actual weld were determined by software.⁴⁸  The measured areas for each weld zone were used to calculate the surface area fraction, giving values of 33%, 33%, 16.5%, and 16.5% for the wrought base, HAZ2, HAZ1, and FZ, respectively. Therefore, 4, 4, 2, and 2 equal surface area electrodes and ZRA channels were used for the weld sections, respectively. A Scribner model MMA910B^† was used to provide a graphical interface and data acquisition of each microelectrode current. This equipment is capable of galvanically coupling and measuring up to 100 working electrode current channels and contains an individual ZRA on each current channel with a measurable current range of 3.3 nA to 100 μA per channel. Neither a counter electrode nor a reference electrode was used in this study as only localized galvanic currents were obtained. Samples were immersed in 0.6 M NaCl, 0.6 M NaCl saturated with Mg(OH)₂, and TRIS, to examine how the corrosion changed with the environment used. Each MMA environment was tested at least three times and a typical run is shown herein.

Raman spectroscopy was performed utilizing a Renishaw InVia^† Raman microscope for AZ31 samples after immersion in 0.6 M NaCl and 0.1 M TRIS environments for 24 h at OCP. Samples were kept in solution until approximately 30 s before Raman spectroscopy was performed on the sample surface to alleviate any issues with air-formed oxide growth. All samples were analyzed utilizing a 514 nm wavelength laser operating at 100 mV under a 50× objective lens with 3,000 L/mm grating. Raman analysis was performed on approximately three spots on the sample surface and a typical Raman spectrum is reported.

The corrosion rate of the isolated weld zones and the wrought base were reported elsewhere.^7,41  However, when intact welds are exposed in industrial service in the presence of an electrolyte, each zone of the weld is connected (i.e., electrically and ionically).

Concerning the intact weld, SEM and EBSD were taken along the edge of the FZ and HAZ in the same location before and after corroding the sample for 3 h at OCP in 0.6 M NaCl, as shown in Figure 3. From the metallurgy of the individual weld zones as summarized in Figure 3, the FZ had a strongly randomized texture commensurate with grains solidified from the melt (Figure 3[b]). Also, a variation in compositional contrast (Figure 3[a]) revealed Al-Zn rich solidification boundaries formed during non-equilibrium processing in the FZ, as reported previously.^7,54  The α-Mg matrix in this zone was depleted of Al and Zn alloying elements as determined by EDS (Table 1). From the corrosion morphology, it was observed that the corrosion propagated in or along the solidification boundaries in the FZ and in proximity to the Al-Mn-Fe IMPs.⁷  The HAZs have their own distinct metallurgical features including a persistent basal texture (Figure 3[b]) originating from the base plate texture combined with grain growth and an α-Mg matrix composition which was similar to the wrought base (Table 1). Many Al-Mn and Al-Mn-Fe particles were detected throughout this zone; their compositions are listed in Table 1. Concerning the corrosion morphology, the HAZ corroded predominantly by the filiform corrosion morphology (Figure 3[c]) along the surface in the direction.

In 0.6 M NaCl solution saturated with Mg(OH)₂, the corrosion morphology in the FZ (Figures 4[a] and [b]) was similar to that seen in 0.6 M NaCl-only solution, with corrosion propagating at the solidification boundaries and proximate to IMPs. In the HAZ1 and HAZ2 (Figures 4[c] through [f]), the filiform morphology was dominantly observed. The filiform filaments grew along the surface of the material. Much of the filiform corrosion initiated near the Al-Mn-Fe IMPs throughout the material and many of these particles remained on the samples surface even after corrosion (circled in orange).

The AZ31B TIG weld was also immersed in a non-chloride environment, 0.1 M TRIS. The FZ (Figures 5[a] and [b]) was rather more uniformly corroded than in the chloride-containing environments. However, this corrosion morphology propagated further than in the case of the chloride environments (Figures 3[c] and 4[b]). In the HAZs (Figures 5[c] through [f]), the corrosion was much more extensive than in the chloride-containing environments (Figures 3[c], 4[d], and 4[f]). In both of these weld zones, the Vickers microhardness indentations were mostly eliminated, presumably as a result of surface recession. However, there was a different pattern or morphology, with slight L directionality along the surface. Also, none of the Al-Mn-Fe particles that were originally on the sample surface were detected after corrosion.

It has been shown in the literature that as the material corrodes, the surface becomes dark⁵⁵  and this region typically contains enriched transition elements^32,56  or replated Al and Al³⁺ deposits.²⁹  Scanning vibrating electrode technique measurements show enhanced cathodic reactions at the edge of these regions where anodic reactions are enhanced.^17,28,55  Dark regions are typical of the anodic Mg dissolution morphology that develops in chloride-containing solutions.⁵⁵  The values found for 0.6 M NaCl are statistically similar in the 0.6 M NaCl saturated with Mg(OH)₂. Figure 6(a) contains the percent of dark surface area for each weld zone in isolation and Figure 6(b) contains the percent surface area for each weld when galvanically connected. The wrought base material has the greatest percentage of this corroded region formed compared to any TIG weld zones, both in the case of isolated weld zones⁷  and for the galvanically connected weld. The FZ has the next highest percent area corroded, compared to the HAZs, for both the isolated and the galvanically connected welds after 24 h. The percent area corroded in the galvanic case is slightly higher for each zone when galvanically connected (Figure 6[b]) compared to the percent area corroded in isolation (Figure 6[a]). This corresponds to the higher corrosion rates observed when the weld zones are electrochemically connected in contrast to the situation when the weld zones are corroded in isolation.⁷ 

The apparent galvanic corrosion density of the isolated weld zones in electrochemical connection to other weld zones was measured over a 24 h period. In 0.6 M NaCl, the FZ was electrically connected to the HAZ1 (Figure 7). The galvanic potential of the FZ and HAZ1 immediately after immersion was −1.52 V_SCE but decreased, after 3 h, to −1.60 V_SCE (Figure 7[a]). Initially, the HAZ1 was the anode and the FZ was the cathode in the galvanic couple between the FZ and HAZ1. However, after 3 h, the polarity between the zones switched and the FZ became the anode (Figure 7[b]). Connecting FZ to the HAZ2 had similar results. The galvanic potential of the FZ and HAZ2 zones were approximately −1.54 V_SCE when initially immersed but decreased to −1.59 V_SCE after 3 h. The HAZ2 was initially the anode but switched polarity after 3 h and the FZ became the anode. In the galvanic couple containing the FZ and the wrought base, the galvanic potential was approximately −1.57 V_SCE initially but increased to −1.53 V_SCE after longer immersion times. The FZ was the anode initially but the wrought base became the anode after 3 h.

In 0.6 M NaCl saturated with Mg(OH)₂, the galvanic potential and current were also measured (Figure 8). The galvanic potential between the FZ and the HAZ1 was −1.55 V_SCE immediately after immersion and then −1.58 V_SCE after 24 h (Figure 8[a]). The HAZ1 initially was the anode but, after 3 h, the polarity reversed and the FZ became the anode (Figure 8[b]). The galvanic couple potential of the FZ and HAZ2 couple was initially −1.44 V_SCE and then decreased to −1.54 V_SCE. The HAZ2 initially functioned as the anode but, after 3 h, the polarity reversed and the FZ became the anode. The galvanic couple potential for the FZ and the wrought base was −1.55 V_SCE at 0 h and −1.51 V_SCE at 24 h. Initially, the FZ was the anode but, after 3 h, the polarity reversed and the wrought base became the anode.

The samples were next immersed in the non-chloride-containing environment, TRIS (Figure 9). No polarity reversal was seen in this environment. The galvanic couple potential of the FZ and HAZ1 was initially −1.59 V_SCE and was −1.39 V_SCE after 24 h (Figure 9[a]). In the FZ to HAZ1 galvanic couple, the HAZ1 was the anode the full 24 h (Figure 9[b]). When FZ was immersed in connection with HAZ2, the galvanic couple potential was measured to be −1.68 V_SCE immediately after immersion and −1.42 V_SCE at 24 h. In the FZ to HAZ2 galvanic couple, the HAZ2 was the anode the full 24 h. The galvanic couple potential for the FZ and wrought base couple was initially −1.68 V_SCE and then −1.42 V_SCE after the 24 h immersion time. In the FZ to wrought base galvanic couple, the wrought base was the anode for the entire 24 h. Hence, polarity switching was not observed generally in TRIS.

The MMA was tested in the same three environments. In 0.6 M NaCl (Figure 10), there was a variation in the current measured with time. Different zones also underwent anodically-induced cathodic activation, which was demonstrated by a polarity reversal in the measured current (Figure 10). The currents switched from initially positive and anodic to negative in a consistent sequence. When the sectioned weld zone initially activated, there was an abrupt increase in the measured positive current, which remained high for a similar time as seen in binary galvanic couple experiments. Then, the activated weld section became slightly less anodic, but remained positive for an additional period of time until the polarity was reversed (the current became negative). The HER was observed to occur dominantly on these anodically activated weld regions during this polarity reversal. The time interval of this cathodic activation was not the same for each weld region. Eventually each anodic weld zone section deactivated and then became a strong cathode. Examining the current, the strongest anode, after deactivation, became a strong cathode. The wrought base and HAZ2 were the most likely to undergo anodically-induced cathodic activation.

Concerning the MMA as exposed in 0.6 M NaCl saturated with Mg(OH)₂, a similar trend was shown (Figure 11). Each zone activated with time, and after anodic activation the channel suffered a polarity shift and became a strong cathode eventually. This cathodic current then slightly decreased. After some period of time, all anodically-induced cathodically activated weld zones became strong canthodes.

A different trend was shown in the MMA exposed in TRIS (Figure 12). Immediately after immersion, one of the wrought base zones started as an anode. Examination of this same channel (green) at the end of the 24 h test revealed it had become a cathode. Similarly, the FZ (gray) started as a strong cathode but ended as an anode. However, the changes were not as significant. While some polarity reversal was observed on various channels, there was much less than previously seen in the chloride environments and less anodically-induced cathodic activation was observed (Figures 10 through 12).

A variation in the oxide formation may alter the corrosion response, including the polarity reversal and resultant anodically-induced cathodic activation. The pH for the chloride-containing environments before exposure was typically around 6.5 before becoming more alkaline with the generation of OH⁻, ~11. However, the pH of TRIS was 7 after exposure, which does not allow for Mg to as readily passivate. This was confirmed through the use of Raman spectroscopy (Figure 13). All scans were taken with the same instrumental parameters so they could be compared fairly. Fitting of the Raman shift to the characteristic Mg(OH)₂ peaks showed a strong presence of the Mg(OH)₂ film in the chloride-containing 0.6 M NaCl solution. However, these peaks had less intensity in the TRIS solution, which suggested that only a small amount of Mg(OH)₂ was formed. This variation in the oxide thickness with exposure environment may lead to a variation in the corrosion response.

From the variation in the OCP with time (Figure 14), the trend of the OCP differed depending on whether the weld zones were isolated or galvanically connected. When the weld zone OCPs were measured in isolation in 0.6 M NaCl (Figure 14[a]), the highest OCP was observed for each weld zone after 3 h, with the FZ having the most noble OCP. This is consistent with large Al-Mn or Al-Mn-Fe particles and Al-Zn solidification structures. A polarity reversal was observed between the weld zones around 12 h coupled when measured in isolation with the OCP in the order of FZ > HAZ > wrought base from 0 h to 12 h and HAZ > FZ > wrought base from 12 h to 24 h. However, when the weld zones were galvanically connected and then disconnected to access OCP (Figure 14[b]), the OCP increased over the 24 h period for each weld zone and the highest OCP was measured at 24 h. The polarity reversal for the galvanically connected weld zones occurred after only 3 h where the OCP trend was FZ > HAZ > wrought base from 0 h to 3 h but HAZ > FZ > wrought base from 3 h to 24 h. These numbers were statistically similar to those achieved when Mg(OH)₂ was added to solution and have been excluded for redundancy.

However, the electrochemical behavior in the non-chloride, TRIS environment differed from the chloride environments (Figures 15[a] and [b]). For both the isolated and galvanically connected weld zones, the measured OCP increased over the full 24 h immersion. The largest increase occurred from when the sample was first immersed up to 3 h, with only a small increase between 3 h and 24 h. Only slight polarity reversal (Figure 12) and little cathodic activation (Figure 15) were seen during the 24 h immersion, and the magnitude of the OCP in order was FZ > HAZ1 > HAZ2 > wrought base.

The OCP as a function of time and weld zone over a 24 h immersion period in a chloride environment is shown in Figure 16(a) for each weld zone after being galvanically connected for 3 h, 24 h, and 48 h and then disconnected to measure OCP and assess cathodic kinetics. When the sample was first immersed, the FZ had the highest potential of each of the weld zones, and OCP increased up to 3 h of immersion for each weld zone. After 3 h, the HAZs had more negative OCPs than the FZ. After 12 h and continuing on to 48 h, the OCP continued to increase for each galvanically connected weld zone. The net current as a function of weld zone was assessed for specified times to examine the current trends with time (determined by the MMA) (Figure 16[b]) and different weld regions were cathodic at various times. Each weld zone activated at different times. To further understand how the current changed at different locations with time, selected exposure times were selected and the cathodic current density (i_c) at −1.8 V_SCE was plotted as a function of weld zone for specified times (Figure 16[c]). i_c increased with time for all weld regions (Figure 16[c]), except for the HAZs, where there was a slight decrease or no change after 24 h. The increase was the greatest for the wrought base and the least for the HAZ2 over time. The large increase in i_c for the base material was consistent with anodically-induced cathodic activation.^25,28  The net current as a function of weld zone for specified times is shown for the 0.6 M NaCl saturated with Mg(OH)₂ as well (Figure 17). It was clear that different weld regions were cathodic at various times. The trend in the cathodic activation for the chloride-containing environment saturated with Mg(OH)₂ was the same as the trend shown in 0.6 M NaCl, and the variation in the OCP and i_c with time is excluded for brevity.

The OCP as a function of time and weld zone over a 24 h immersion period in a non-chloride environment was shown (Figure 18[a]) for each weld zone after being galvanically connected for 3 h, 24 h, and 48 h and then disconnected to assess cathodic kinetics. It was observed that the OCP slightly ennobled for each weld zone with time and that the FZ was the most cathodic at each time step. Figure 18(b) illustrates that different weld regions were cathodes at various times. It was also shown that, unlike in the chloride environments, multiple channels were cathodic at the same time. The variation in the net current was smaller than in the chloride environments. The i_c at −1.8 V_SCE was plotted as a function of time in each of the isolated weld zones after 3 h, 24 h, and 48 h in 0.6 M NaCl. In Figure 18(c), there was a slight increase in the i_c with time; however, this increase was much smaller than observed in the chloride environments (Figure 16). The FZ had the lowest i_c, while the wrought base had the highest after each exposure time.

From the resultant corrosion morphology characterized through SEM imaging, several corrosion morphologies were seen in a weld. These corrosion morphologies were dependent on the weld zone and exposure environment. In the chloride environments, specifically 0.6 M NaCl and 0.6 M NaCl saturated with Mg(OH)₂ (Figures 3 and 4), the corrosion propagated along the Al-Zn solidification boundaries and proximate to Al-Mn-Fe particles within the FZ. These solidification boundaries were rich in the alloying elements and cathodic to the α-Mg (which had been depleted in Al and Zn [Table 1]).⁴  The galvanic couple formed between the α-Mg and the solidification structure caused the Mg directly adjacent to this solidification structure to corrode along with the α-Mg (Figures 3[c] and 4[b]). Similarly, the cathodic Al-Mn-Fe particles were affected during corrosion (Figures 3[c] and 4[b]). In the HAZ, the dominant corrosion morphology was filiform growth (Figures 3[c], 4[d], and 4[f]). Al-Mn-Fe particles were distributed throughout the material. Filiform corrosion propagated in the proximity of these Al-Mn-Fe particles, as has been reported in the AZ31 material^54,57-58  and other Mg alloys^59-63  (Figures 3[c], 4[d], and 4[f]). This corrosion attack did not appear to be isolated to one grain, but rather propagated through several grains.

In TRIS (Figure 5), the corrosion morphology was drastically different than the corrosion morphology seen in 0.6 M NaCl. In the FZ, the Al-Zn solidification boundaries and Al-Mn-Fe particles preferentially corroded as a result of a similar galvanic mechanism as in the chloride environment (Figures 5[a] and [b]). The directionality of the corrosion propagation was slightly evident along the surface in the HAZs; however, the morphology uniformly covered the surface (Figures 5[c] through [f]). The Al-Mn-Fe particles were not detected on the sample surface. In a TRIS environment, only a thin Mg(OH)₂ film was formed during corrosion as determined from the weak Mg(OH)₂ Raman peaks (Figure 13). As a result of the thin Mg(OH)₂ film, the corrosion rate was higher with more uniform corrosion on all zones and phases.

There are small differences in the OCP between each of the weld zones (Figures 14 and 15). This will create a galvanic couple between the weld regions and increase the corrosion rate. These differences in the measured OCP are at first a result of the microstructural and metallurgical features such as grain size, crystallographic orientation, solidification structures, and IMPs. The electrochemically connected weld zones then interact with one another when immersed in the same solution for an extended period of time. Polarity reversals then start, which tend to dominate the galvanic behavior instead of the initial microstructure. Because of this, cathodic activation is a key factor as it controls the polarity and galvanic couple relation (i.e., anode or cathode). These have all been factors in Mg corrosion and are the subject of ongoing studies.

The anodically-induced cathodic kinetics as a result of AZ31B TIG welding varied depending on environment. In chloride environments, such as 0.6 M NaCl, cathodic activation occurred in all weld zones and led to polarity reversal during galvanic coupling (Figures 7 and 8). This cathodic activation was monitored using a MMA to examine initial anodes and subsequently revealed which areas of the weld cathodically activated with time (Figures 10 through 12). In the chloride environments, individual weld sections functioned as anodes based on their starting OCP relative to the galvanic couple potential. However, these initial anode zones eventually were cathodically activated and then became strong cathodes. Most of the individual weld sections that became a strong anode subsequently became a strong cathode and the polarity was reversed during galvanic corrosion (Figures 10 and 11). Therefore, copious HER was seen to occur on those weld sections that became cathodically activated. In the TRIS environment, a large cathodic activation process was not observed (Figures 12 and 15). The HER occurred more uniformly across all of the weld sections. However, i_c increased somewhat when anode zones were interrogated (Figure 18).

Considering the increase in i_c with time (Figure 16[c]) in 0.6 M NaCl, it was clear that cathodic activation may occur in the base material, as well as in the weld zones, as determined by an increase in the cathodic kinetics and OCP with time. The variation in the i_c with time in TRIS (Figure 18[c]) was much smaller than in the chloride environments. This supports the observation that there is less cathodic activation in the non-chloride, TRIS environment.

The dark regions that expanded over the sample surface (Figure 6) correlated to the anodic Mg dissolution reaction, showed a similar correlation with corrosion rate, and were observed in the chloride environments. The regions of the weld with the highest area fraction of dark region had the highest corrosion rates as determined previously.⁷  Within the dark region, it has been hypothesized that metals other than Mg, which typically remain as impurities within the sample, are not dissolved during the dissolution of Mg and are collected on the partially protective surface film^{27-28,30,54,63}  or replated.^64-65  Cathodic activation has been cited to frequently occur in both commercially pure Mg and Mg alloys such as AZ31²⁵  and has been determined using a variety of approaches.^26,32  It was typically stated that, over time in a commercially pure Mg sample, alloying elements or impurities will enrich to the sample surface while, in a Mg-Al alloy, cathodic activation can cause the enrichment of Al-Mn phases on the sample surface in the partially protective surface film.^{27-28,30,54,63}  This enrichment^{26-28,30,54,63}  or replating^64-65  will ennoble the sample surface and will increase the cathodic reaction rate over time. In response to this enrichment or replating at the sample surface, the i_c will increase as well as cause the polarity reversal in weld zones. This process is an important factor in weld corrosion where the initial microstructure is important to the establishment of initial anodes and cathodes but the ability to undergo anodically-induced cathodic activation plays a role in weld corrosion, even in 24 h.

The variation in the film formation with environment may alter the corrosion rate and weld corrosion behavior. Samples with an Mg(OH)₂ film have a greater difference in OCP within each weld zone and a greater cathodic reaction rate. This finding is particularly evident in environments that allow a large amount of Mg(OH)₂ formation. Samples immersed in chloride-containing environments develop a thick Mg(OH)₂ layer while, in TRIS, only a thin film may be developed, as shown through Raman spectroscopy (Figure 13). It was observed that the polarity reversals and cathodic activation in TRIS were less extreme, which may be a result of the variation of this film formation.

• The corrosion of an AZ31B-H24 TIG weld varied with time and was dependent on the galvanic corrosion effects between electrochemically connected weld zones. All chloride environments functioned similarly because of excess Mg(OH)₂ formation either preloaded or as a result of Mg²⁺ release. TRIS was dissimilar because of suppression of the thick surface film.

• The cathodic activation of individual weld sections was dependent on environment where NaCl and NaCl saturated with Mg(OH)₂ showed large amounts of cathodic activation. However, in a non-chloride environment, TRIS, less of this cathodic activation process was observed. From this study, it was demonstrated that propensity for cathodic activation of each weld zone dominates the corrosion behavior over the long term compared to the metallurgy of each weld zone, which was the dominant factor at early times.

• The OCP and cathodic kinetics, as a function of time, of a galvanically connected weld zone increased more with time than the OCP or cathodic kinetics of a single weld zone in isolation.

This work was funded by the Office of Naval Research Grant N000141210967 with Dr. David A. Shifler as scientific officer. The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the Office of Naval Research and the Technical Corrosion Collaboration.