July 18, 2008

Characterization and Improvement in the Corrosion Performance of a Hot-Chamber Diecast Mg Alloy Thin Plate By the Removal of Interdendritic Phases at the Die Chill Layer

By Uan, Jun-Yen Li, Ching-Fei; Yu, Bing-Lung

This work presents results concerning the effect of the die chill layer on the corrosion performance of a hot-chamber diecast AZ91D thin plate, with particular attention to the role of interdendritic phases (primary beta (Al^sub 17^Mg^sub 12^) and surrounding alpha phase). The primary beta phase in the die chill layer was removed by Ar+ etching, but the other main constituent phases remained on the surface of the sample. Although previous researchers have attributed intense galvanic corrosion to the cathodic phase (primary beta), this work demonstrates that the removal of the primary beta phase from the diecast sample surface did not improve the corrosion performance of the sample. Transmission electron microscope (TEM) examinations were performed, to elucidate the microstructure near the die chill surface, especially focusing on the alpha-Mg phase that surrounds the primary beta phase. The surrounding alpha-Mg phase (known as "Al-rich alpha" or eutectic alpha) actually did not contain a high concentration of aluminum solid solution. Instead, the Al-rich alpha was composed of fine Al^sub 12^Mg^sub 17^ beta particles and Al-Mn-like particles (smaller than 0.5 [mu]m) that were distributed in a low-Al-containing Mg matrix (~4 wt pct Al). Such fine cathodic particles seem to participate strongly in the corrosion. Removing the primary beta phase alone did not increase the corrosion resistance of the material, because many of the cathodic fine particles remained in the Al-rich alpha phase region. This work studied an HF-H^sub 2^SO^sub 4^/CaCO^sub 3^ etching method for removing interdendritic phases (both the Al-rich alpha and the primary beta) from the surface chill layer of the diecast thin plate. Therefore, testing in 5 wt pct chloride solution demonstrated that the I^sub corr^ of the HF-H^sub 2^SO^sub 4^/CaCO^sub 3^- treated specimen was 3 to ~16 [mu]A/cm^sup 2^, which was significantly lower than the I^sub corr^ ~ 1600 [mu]A/cm^sup 2^ of the as-diecast Mg sample. DOI : 10.1007/s11661-007-9420-2

(c) The Minerals, Metals & Materials Society and ASM International 2008


MAGNESIUM alloys are increasingly used for automotive applications because of their capacity to meet vehicle performance requirements.[1] Diecasting is one of the most extensively used approaches for producing magnesium alloy parts. Two basic diecasting processes-hot chamber and cold chamber-are used.[2] The former process has some advantages over the latter, including rapid cycling, improved fluidity, the use of a lower injection pressure, etc.[2] The former process, therefore, offers a favorable method for the mass production of electronic/appliance housings with thin sections. However, a comprehensive microstructural study and the dependence of the microstructure on the improved corrosion resistance of a hot-chamber diecast thin plate (1.4-mm thick, in this work) have been addressed briefly.[3,4]

The Mg alloy exhibits high electrochemical activity,[5] leading to poor corrosion resistance, such as in a chloride environment. Chemical conversion is an important surface pretreatment for improving the corrosion resistance of the Mg alloy.[2] Small cracks are distributed on the conversion coating layer during dehydration, improving the adhesion of subsequent paint layers or organic coatings to the surface of the Mg alloy substrate. The conversion coating on magnesium has been traditionally based on hexavalent chromium compounds.[2] However, the hexavalent chromium (Cr^sup 6+^) is a toxic substance that pollutes the environment and detrimentally affects health.[6] Recently, many investigations have studied an alternative, chrome-free conversion coating process to protect magnesium against corrosion.[7-11] In the new conversion treatments, the Mg sample was immersed in a bath of permanganatephosphate (KMnO^sub 4^ + MnHPO^sub 4^),[9,10] stannate (NaOH + K^sub 2^SnO^sub 3^ . 3H^sub 2^O + NaC^sub 2^H^sub 3^O^sub 2^ . 3H^sub 2^O + Na^sub 4^P^sub 2^P^sub 7^),[12,13] cerium nitrate (Ce(NO^sub 3^)^sub 3^),[8,11] CeCl^sub 3^/ H^sub 2^O^sub 2^,[14] La(NO3)^sub 3^,[8] Pr(NO3)^sub 3^,[8] or cobalt-III hex coordinated complex (Co(NO^sub 3^)^sub 2^ + NH^sub 4^NO^sub 3^ + NH^sub 4^OH + NH^sub 3^).[15] The Mg sample that was treated in the various aforementioned baths is coated with various compounds. For example, the conversion coating films may be composed of Mg^sub 3^(PO^sub 4^)^sub 2^/MgMn^sub 2^O^sub 8^,[9] MgSnO^sub 3^,[13] cerium oxide/hydroxide,[8,11,14] lanthanum oxide/hydroxide,[8] or praseodymium oxide/hydroxide.[8] The cited investigations have contributed substantially to the success of the Cr^sup 6+^-free conversion coating process. However, with respect to recycling magnesium scraps from automotive components or any postconsumed Mg product (end-of-life goods), applications of chemical conversion coatings make difficult the recycling of the scraps into high-quality diecasting alloy ingots that meet ASTM specifications.[16,17] Some of the main reasons are that the magnesium melt becomes contaminated by surface contamination and the formation of dross increases because of the presence of the conversion layer.[17] According to Skar et al.,[18] low recyclability and high toxicity are associated with the formation of conversion films on the surface. Meskers el al.[19] studied the thermal decoating of Mg alloy scraps (end-of-life products made of Mg alloys), finding that this step is an initial step toward the recycling of coated Mg. Recently, some studies have attempted to coat a Mg-friendly metal on magnesium alloy to protect against corrosion. For instance, Yamamoto et al.[20] coated a thick, pure magnesium film on a Mg-Al-Zn alloy sample surface by vapor deposition. Their work concluded that a pure magnesium film may not only improve corrosion resistance but also increase the recyclability of alloys. Yu and Uan[21] coated a finegrain Mg thin film on an AZ91D surface by vapor deposition. The film served as a sacrificial anode and cathodically protected the cathode (the AZ91D substrate). However, metallurgical factors that can improve the corrosion performance of diecast Mg alloys include the microstructure[22-24] and the chemical impurities.[25,26] For example, Song et al.[24] examined the corrosion characteristics of a cold-chamber diecast AZ91D alloy. The as-cast sample had a thickness of 6 mm, and some microstructural discrepancies were present between the die skin layer and the interior of the sample. Song et al. inferred that an as-cast structure on the 6-mm-thick sample surface with a fine cast microstructure and a high fraction of networklike Al^sub 12^Mg^sub 17^ beta phase has better corrosion resistance than a sample with a coarser microstructure. Blawert et al.[25,26] found that the skin of various cold-chamber high-pressure diecast (HPDC) Mg alloys had more inferior corrosion properties than did the bulk. They explained that the poor corrosion performance of the skin was caused, in part, by contaminating Fe, Ni, and Cu, which were likely from the die casting mold and die lubricant.[25] Therefore, according to Blawert et al., simply removing the outermost skin by grinding clearly increases the corrosion resistance of the HPDC alloys. The preceding experimental results may have important implications for improving the corrosion resistance of the diecast Mg alloy without chemical conversion coating. However, with regard to the die skin microstructure of hotchamber diecast thin plate, rapid cooling of the diecast for a thin gage section yields nonequilibrium solidification structures that comprise a large fraction of the Al-rich alpha phase and a relatively low fraction of Al^sub 12^Mg^sub 17^ beta particles in the die skin of a thin diecast plate.[27] These findings differ from those for a thick diecast.[24] A study by Yu and Uan[27] demonstrated that a hot- chamber diecast thin plate (1.4-mm thick) with die skin on its surface was severely corroded in chloride solution. Therefore, concerning a hot-chamber diecast plate with a thin gage section, questions include why the thin diecast plate with die skin on the surface exhibits poor corrosion resistance and how to improve the corrosion performance of the thin diecast without an extra chemical conversion coating or mechanical surface grinding. This work elucidates the phases in the die skin that were mainly responsible for the inferior corrosion performance of the thin diecast plate. Removing the corrosive phases from the die skin markedly improved the corrosion resistance of the thin diecast Mg alloy plate. The micro structures, the corrosion behavior of the die skin, and the process that removed some specific phases from the die skin were studied in detail herein.


A hot-chamber diecast panel for a notebook computer (Figure 1) was adopted herein. The diecast thin plate used in this work was the same as that used in the authors' earlier studies.[4,21,27] The panels are qualified, and the surfaces have limited casting defects. Auger electron spectroscopic (AES) analysis and the AES compositional depth profile were employed to evaluate the compositions on the skin surface of the as-diecast AZ91D panel,[27] revealing neither surface pollutants nor the segregation of Fe, Ni, and Cu impurities. hot-chamber diecasting processing, with details of the casting parameters that yielded the thin panel used herein, was described in the authors' previous study.w The upper side of the panel (Figure 1) was exposed to be etched and to be tested in a corrosive environment. Samples sized 20 x 20 x 1.4 mm^sup 3^ were cut from the panels. All samples were from the rectangular region on the panel (Figure 1), to minimize the possible discrepancies among the chemical compositions and microstructures of the samples. Table I presents the chemical composition of the alloy, which meets the ASTM B94 standard specifications of the AZ91D alloy. Table II presents solutions for selected etching treatment. Sulfuric acid (H^sub 2^SO^sub 4^) and hydrofluoric acid (HF) were used to prepare the etchants. These acids meet the referential of American Chemical Society specifications (ACS grade).[28] Samples were immersed in the first solution (Table II) for 3.5 minutes. After those samples had been etched, they were cleaned in distilled water. They were then dipped in the second solution (Table II) for 3 minutes, before being cleaning in distilled water. Each sample surface turned black following the selected etching treatment. Some of the samples were further immersed in the thin slurry that contained 5wt pct CaCO^sub 3^ in 100 mL of distilled water, and then were rinsed in distilled water. The purpose of the CaCO^sub 3^ treatment was to damage slightly the black surface film on the selected etching sample, if a film existed. Therefore, subsequent corrosion tests helped to determine whether the black film protected against corrosion.

To elucidate the role of the primary beta phase in the die chill layer in the corrosion resistance of the diecast thin plate, physical etching was performed, to remove the primary beta phase from the surface of the die skin. The use of different equipment for ion etching may produce various results. Hoche et al.[29] adopted a physical vapor deposition (PVD) system to perform sputter cleaning of the AZ91D sample before PVD deposition. Surface observations after 200 minutes of sputter cleaning demonstrate that the alpha phase is etched with a rough surface, while Al^sub 12^Mg^sub 17^ beta particles remain. The intermetallic Al^sub 12^Mg^sub 17^ is well known to be brittle, but the alpha-Mg phase is more ductile. According to Shin et al.[30] and Wellman et al.,[31] the rate of removal of brittle material by high-speed erodent reaches its maximum when the impact angle of the erodents to the target surface is about 90 deg, while the 90-deg impact angle results in a relatively low rate of removal of a ductile phase. Therefore, a Gatan precision etching and coating system (PECS, Gatan, Inc., Pleasanton, CA) was employed herein. An argon ions beam generated via the PECS had a normal incident angle relative to the surface of the specimen, with a distance of only 58 mm between the ion gun tip and the surface of the sample. The ion beam energy applied herein was 6 Kev, with ion current density 250 [mu]m/cm^sup 2^. The optimal ion etching time was 60 minutes. The etching time was one of the most important parameters. Proper control of the period of etching by the Ar+ ions resulted in the successful removal of Al^sub 12^Mg^sub 17^ beta particles, but most of the Al-rich alpha phase remains on the as-cast die skin.

A salt spray test and electrochemical polarization experiments were performed, to evaluate the corrosion resistance of the samples. An ASTM B117[32] salt spray test was conducted, to determine the corrosion performance of the samples. The samples were kept in a chamber with salt spray at 35 [degrees]C for 168 hours. A 5 wt pet NaCl aqueous solution was used in the tests. At the end of the experiments, the samples were cleaned by dipping in a solution of 15 wt pet CrO^sub 3^ + 1 wt pct AgCrO^sub 4^ in 100 mL of boiling water. The corrosion rate was determined in millimeters per year (mmpy). The corrosion data were collected from four samples under each testing condition. Electrochemical polarization tests were conducted in a corrosion cell that contained 270 mL of a 5 wt pet NaCl solution at room temperature, at a scan rate of 0.5 mVs^sup - 1^. All electrochemical measurements were made using a Princeton Applied Research EG&G Model 263A* potentiostat/galvanostat and EG&G M352 software**. The surface area that was exposed to the solution was 1 cm^sup 2^. A platinum gauze was adopted as a counter electrode, and silver chloride was used as a reference. In each experimental case, at least three experiments were performed. The electrochemical impedance spectra following selected etching treatment and those of the as-diecast specimens were also measured, in 5 wt pct NaCl solution at 25 [degrees]C, using a Princeton Applied Research EG&G Model 5210[dagger] lock-in amplifier electrochemical measurement system. The amplitude of the applied alternating current (AC) signal was 5 mV, and the range of measured frequencies was from 10 mHz to 100 kHz.

* EG&G Model 263A is a trademark of AMETEK Princeton Applied Research, Oak Ridge, TN.

** EG&G M352 software is a trademark of AMETEK Princeton Applied Research, Oak Ridge, TN.

[dagger] EG&G Model 5210 is a trademark of AMETEK Princeton Applied Research, Oak Ridge, TN.

The microstructure and elemental composition were studied using a JEOL[double dagger] JSM-6700F field-emission scanning electron microscope (FE-SEM) equipped with an Oxford Inca Energy 400 energy- dispersive spectrometer (EDS, Oxford Instruments, UK). Quantitative EDS analysis was performed at an accelerating voltage of 12 kV. Correct analytical treatment of the EDS data requires recording spectra from a known standard sample for calibration. Pure copper with a well-polished surface was the standard sampled[27] The crystallographic structure was studied by grazing-angle X-ray diffraction (GAXRD), with an angle of incidence of 1 deg and a scanning rate of 0.5 deg/min.

[double dagger] JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.

An FEI Tecnai G2 F20 transmission electron microscope (TEM, FEI Company, Hillsboro, OR) equipped with a Z-contrast imaging system was employed to examine the microstructure in die skin. This Z- contrast imaging technique supports atomically resolved compositional sensitivity in the TEM. The signal intensity on the image is proportional to the average atomic number, and therefore, the Z-contrast.[33] Elemental microanalysis of individual entities was performed using TEM/ EDS with a beam size of 10 nm. The TEM samples were prepared by mechanical polishing from the bottom of the sample, to retain the die skin. The samples were polished to a thickness of ~70 [mu]m. They were punched into 3-mm-diameter discs, which were then slightly polished to a thickness of ~20 [mu]m. Polishing was followed by ion milling, thinning of the discs until a hole formed with thin foil edge on the discs. Since the Mg alloy sample adopted herein was the same as that used in our earlier work,[27] the thickness of the die skin was determined, and was 20 to ~35 [mu]m.


A. Surface Microstructure

Figures 2(a) and (b) present surface SEM images of the diecast sample that reveal the die skin structure. The microstructure was revealed by slight polishing and minor chemical etching. As illustrated in Figure 2, the Al-rich alpha phases appeared dark, and the primary Al^sub 12^Mg^sub 17^ beta phase appeared bright. The rest of the die skin structure was alpha-Mg phase, as shown in Figure 2 in gray. The Al-rich alpha phase exhibited a networklike morphology, located at interdendritic spacing, as displayed in Figure 2. Under high magnification (Figure 2(b)), the primary Al^sub 12^Mg^sub 17^ beta compound was a particle-like phase with an irregular shape. The beta phase was embedded in the Al-rich alpha phase, as displayed in Figure 2. A table in Figure 2 indicates that the Al concentration of the Al-rich alpha phase (position 1) was approximately 13.55 wt pet of Al, as determined by FE-SEM/EDS. For comparison, the alpha-Mg phase (position 2) contained ~3.92 wt pet Al. The die skin on the thin diecast plate was composed of primary alpha-Mg, Al-rich alpha-Mg, and primary Al^sub 12^Mg^sub 17^ beta phase. The FE-SEM/ EDS line scanning was performed, to observe the microstructure of the die skin and the concentration profiles among various phases. Figure 3(a) shows a surface FE-SEM image of the die skin. Figure 3(b) presents the concentration profile of the elements along the line displayed in Figure 3(a). In Figure 3, the primary alpha-Mg grain, the Al-rich alpha phase, and the primary beta phase corresponded to positions A, B (D), and C, respectively. The primary beta phase (position C) and the Al-rich alpha phase (positions B and D) was composed of a high percentage of Al. In a primary alpha-Mg grain (e.g., position A), the Al concentration fell to ~4.5 wt pet.

The FE-SEM photographs in Figure 4 present the typical surface morphology of the sample following selected etching treatment. The surface of the sample following the selected etching was black. Figures 4(a) and (b) indicate that the interdendritic spacing was preferentially etched, revealing fine as-cast microstructure. Figure 4(b) presents the selected etching microstructure under higher magnification, indicating that the interdendritic spacing was carved more deeply by the etchant than were the alpha-Mg grains. The primary Al^sub 12^Mg^sub 17^ beta phase, which was previously shown as bright (Figure 2), was not seen on the etched surface. Positions 1 and 2 are indicated by arrows in Figure 4(b). A table in Figure 4 presents the Al and F concentrations at positions 1 and 2. Those at position 1 (the surface of alpha-Mg grain) were ~5.16 wt pet Al and ~4.85 wt pct F, while those at position 2 (the interdendritic position) had low Al contents (~3.84 wt pct) and relatively high F contents (~21.59 wt pct F). The Al-rich alpha phase was no longer present in the interdendritic spacing after the selected etching treatment. Moreover, the Mg alloy sample surface turned black following the etching treatment, which might be due to the F on the sample surface. Some of the etched samples were dipped in a thin slurry of CaCO^sub 3^/H^sub 2^O mixing for a short period, to damage slightly the black surface film. In fact, CaCO^sub 3^ treatment turned some of the black surface bright. Figure 4(c) presents a typical micrograph of the sample following selected etching and CaCO^sub 3^ treatment. The table presents the FE-SEM/EDS results for the arrowed positions 1' and 2'. As shown, the F content was not detected for either position. The goal of the CaCO^sub 3^ treatment was to damage slightly the black film on the selected etching sample, if the film is present. Thus, subsequent corrosion tests helped to determine whether the black film protected against corrosion in chloride solution. Figure 5 presents the GAXRD spectra of the as-diecast sample, the selected etching sample, and the selected etching/CaCO^sub 3^-treatment sample, for comparison. The as-diecast spectrum included peaks of the alpha-Mg and Al^sub 12^Mg^sub 17^ beta phases. The peaks of Al^sub 12^Mg^sub 17^ beta almost disappeared after the sample had undergone selected etching treatment. The GAXRD spectra of the selected etching sample and the selected etching/CaCO^sub 3^ treatment sample were consistent with the microstructure following selected etching treatment (Figure 4), which did not show a primary beta phase on the etched surface. Additionally, Figure 5 presents the two-theta positions of each possible fluoride compound. No X-ray peak of a fluoride was observed.

B. Sail Spray Test and Electrochemical Analysis

Figure 6 plots the corrosion rates of the specimens that were statically held in a 5 wt pct NaCl spray chamber for 168 hours. Three specimens were tested; they were as-diecast, the selected etching sample, and the sample that was being treated with selected etching plus CaCO^sub 3^ treatment. As stated, the as-diecast sample had higher corrosion rates than the samples with etched surfaces. The CaCO^sub 3^ treatment, which was performed after the sample had been treated by selected etching, retained good corrosion performance.

Electrochemical polarization experiments were performed in 5 wt pct NaCl aqueous solution, to determine the electrochemical properties of the samples before and after surface treatment. Figure 7(a) plots typical polarization curves in the 5 wt pct NaCl solution. As displayed in Figure 7(a), the as-diecast sample had a higher cathode current density than the samples that had been etched and had undergone CaCO^sub 3^ treatments. Additionally, the corrosion potential (E^sub corr^) of the as-diecast sample was approximately at ~1.55 V^sub Ag/AgCl^, while the selected etched/ CaCO^sub 3^ samples had averaged E^sub corr^ values of as high as about ~1.45 V^sub Ag/AgCl^. Figure 7(b) plots the corrosion current densities (E^sub corr^) of the samples. The E^sub corr^ values of the as-diecast samples ranged between 500 and 2000 [mu]A/cm^sup 2^, while that of the selected etched samples ranged between 10 and 16 [mu]A/cm^sup 2^. About 9 [mu]A/cm^sup 2^ was obtained from the samples that were treated by the selected etching/ CaCO^sub 3^ process. The results of the salt spray tests (Figure 6) and electrochemical polarization scanning (Figure 7) reveal good qualitative agreement between the two methods, suggesting that selected etching and selected etching/CaCO^sub 3^ treatments greatly increased the corrosion resistance of the diecast sample. The difference in the corrosion resistance between the selected etching treatment and the as-diecast AZ91D is further verified by AC impedance measurements (Figures 8(a) and (b)). The diameter of the capacitive loop in the Nyquist plane represents the polarization resistance of the electrode. A greater polarization resistance typically means a lower corrosion rate. In Figure 8(a), the capacitive loop of the selected etching sample markedly exceeds that of the as-diecast AZ91D. Figure 8(b) shows the Bode plots of the selected etching treatment and as-diecast AZ91D immersion in 5 wt pct NaCl solution. Curve-fitting results demonstrated that the sample after selected etching treatment exhibited a high polarization electrical resistance of ~7.73 x 10^sup 3^ Omega, which exceeded the corrosion resistance of the as-diecast AZ91D alloy (8.51 x 10^sup 1^ Omega). This result is in good agreement with the corrosion results that are presented in Figures 6 and 7.

C. Chemical Composition of Inter dendritic Phase at Die Skin and the TEM Analysis

Figures 9(a) and (b) present a TEM bright-field image and a Z- contrast image of an interdendritic phase region. Figure 9(a) depicts a primary beta phase. The upper right part of Figure 9(a) presents the diffraction pattern of the phase. The Z-contrast image (Figure 9(b)) has several bright spots, indicating that these particles are composed of some chemical elements with high atomic numbers. The chemical compositions of the primary beta phase (position I in Figure 9(b)) and some of the particles (positions 2 and 3 in Figure 9(b)) were determined using TEM-EDS. Table III presents the chemical compositions, indicating the Mg, Al, Zn, Mn, and Fe concentrations at corresponding positions. At arrowed position 1 was the primary beta phase, with 58.30 wt pct Mg, 37.89 wt pct Al, 3.82 wt pct Zn, and undetectable Fe and Mn. Particles at positions 2 and 3 had high concentrations of Mn, as well as a high Al content. Based on the TEM results (Figure 9 and Table III), Al- Mn-Hke particles are present around the region close to the primary beta phase. A bright-field micrograph, presented in Figure 10(a), is a typical image that depicts a primary beta phase (the big blocky one) and the microstructure of the region close to the primary phase. The diffraction pattern of the corresponding blocky phase (Figure 10(b)) suggests the crystal structure of the Al^sub 12^Mg^sub 17^ beta phase. Many fine particles were distributed in the Mg matrix near the blocky beta phase, as presented in Figure 10(a). In fact, some other particles were recognized as the sample was tilted, by improving their contrast (such as in the positions circled with dotted lines). Figures 10(c) to (g) display the diffraction patterns of corresponding particles. With the exception of particle (e) (the corresponding pattern of which is presented in Figure 10(e)), most of the particles were Al^sub 12^Mg^sub 17^ beta (Figures 10(b), (c), (d), (f), and (g)). Particle (e) was an Al-Mn- like particle, with approximately 45.48 wt pct Mg, 35.01 wt pct Al, and 19.51 wt pct Mn, according to TEM-EDS. A table in Figure 10(a) presents the Mg, Al, and Zn concentrations obtained by TEM-EDS at the position that was indicated by the arrow in Figure 10(a). The arrowed position is the alpha-Mg that is close to the primary beta phase. As shown in the table, the position had a relatively low Al concentration-approximately 4.03 wt pct.

D. Ion Etching Removal of Primary Al^sub 12^Mg^sub 17^ beta Phase and the Ion-Etched Sample's Electrochemical Properties

As presented in Figure 11 (a), the surface microstructure of the die skin layer following ion etching treatment exhibits irregular cavities. Figure 11(b) presents typical results of GAXRD analysis of the AZ91D surface layer following the ion etching treatment. Additionally, Figure 11(b) displays the GAXRD spectrum of the as- diecast sample. The GAXRD spectrum of the as-diecast skin included peaks of alpha-Mg and the primary Al^sub 12^Mg^sub 17^ beta phase. The intensity peaks of the primary Al^sub 12^Mg^sub 17^ beta had almost disappeared from the die skin of the as-diecast sample after treatment by Ar+ ion etching. The X-ray result (Figure 11(b)) and the ion-etched microstructural observations (as shown in Figure 11(a)) were consistent, proving that the primary beta particles on the die chill skin was removed by the ion etching process adopted herein. Figure 11(c) displays the concentration profiles of Mg and Al along the EDS analysis line, on the SEM micrograph in Figure 11(c). The Mg concentration profile indicated that the positions close to the cavity (such as positions B and D in Figure 11(c)) had a lower Mg concentration than the other positions. Additionally, the Al concentration profile demonstrated that the positions near the cavity (positions B and D) had ~12 wt pct Al, which evidently exceeded the Al concentration at the other position (e.g., position A). Position C, where the primary Al^sub 12^Mg^sub 17^ beta phase had been originally located, had a relatively low Al content. These results indicated that the Al-rich alpha phase (such as at positions B and D in Figure 11(c)) remained after ion etching process. Figure 12 plots the typical polarization curve of the ion-etched sample in 5 wt pct NaCl aqueous solution. For comparison, the curves in Figure 7(a) were also plotted in Figure 12. The polarization curve of the ion-etched sample shifted a little leftward and upward from that of the as-diecast sample. The mean I^sub corr^ value of the ion-etched samples was approximately 1200 [mu]A/cm^sup 2^, which was slightly better than that of the as-diecast samples, which was approximately 1600 [mu]A/cm^sup 2^. The polarization curves (Figure 7 or 12) of the samples the surfaces of which had been treated by removing all the interdendritic phases were remarkably on the low-I^sub corr^ side. These comparisons and microstructural characterization indicated that the corrosion rate (I^sub corr^) was not improved when only the primary beta phase was removed from the die skin layer. When all of the interdendritic phases were removed from the interdendritic spacing in the die skin layer, the corrosion performance of the AZ91D diecast thin plate was substantially improved (Figures 6 through 8). IV. DISCUSSION

The authors' previous study[27] demonstrated that a hot-chamber diecast thin plate (the same material as used herein) has a die chill layer that is composed of interdendritic Al-rich alpha phase/ primary Al^sub 12^Mg^sub 17^ beta phase/alpha-Mg grains' composite microstructures. (Figures 2 and 3 provide detailed microstructural observations.) The primary beta were irregularly shaped (but not networked)-each embedded in the interdendritic Al-rich alpha phase (Figures 2 and 3). Regardless of the die skin, the interior of the diecast plate had the microstructure of primary alpha-Mg grains surrounded by the interdendritic network of the Al^sub 12^Mg^sub 17^ beta phase.[27] The microstructure of the die skin layer differs markedly from that of the interior of the diecast thin plate. In this work, the selected etching treatment was to modify the microstructures of the diecast chill layer by preferentially removing the phases from the interdendritic region-mainly the Al- rich alpha phase and the primary Al^sub 12^Mg^sub 17^ beta phase (Figures 2 and 3). The interdendritic regions became empty spaces (Figure 4) following the etching treatment. The SEM/EDS examinations (Figure 4) and glancing-angle X-ray analysis (Figure 5) indicated that a single phase (alpha-Mg) was retained in the surface layer of the diecast thin plate. Evidently, the selected etching sample exhibited greater corrosion resistance than the untreated (as- diecast) sample, in both the salt spray test (Figure 6) and the electrochemical test (Figures 7 and 8). Therefore, removing the phases from the interdendritic spacing at the die skin markedly improves the corrosion performance of an AZ91D diecast alloy.

Blawert et al.[25,26] investigated the corrosion properties of the die skin of a cold-chamber HPDC AZ91D alloy. The diecast sample with the original as-cast skin had a poorer corrosion performance than the sample without its original die skin. The result by Blawert et al. was also obtained in present work and in the authors' earlier work.[27J According to Blawert et al., the poor performance of the skin on cold-chamber diecast is caused by contamination with Fe, Ni, and Cu, probably from the casting mold and the die lubricant. The main differences between the hot-chamber diecast process and the cold-chamber process include the melt injection pressure and the pressure during solidification.[2,34] Typically, the injection pressures in the hot-chamber process are from 15 to 40 MPa, while the injection pressures in the cold-chamber process range from 30 up to 100 MPa[34] (40 MPa used in Blawert et al.). The maximum pressure during solidification in the hotchamber machine is 25 MPa, while the pressure during solidification in the cold-chamber machine may reach 120 MPa.[34] The use of high pressure in the cold-chamber process may result in a relatively strong impact between the diecast metal and the mold surface. Therefore, contamination of the cold-chamber diecast skin by Fe, Ni, and Cu probably occurs. Blawert et al. found that removing the diecast skin markedly improves corrosion resistance, by removing the Fe and Cu contamination. A hot-chamber diecast thin plate was used as the test material herein. This work and the authors' earlier study reveal that the hot-chamber diecast sample with the die skin exhibited low corrosion resistance. Although the experimental results are similar to those of Blawert et al., AES surface scanning and AES compositional depth profile analysis in the authors' earlier work demonstrated an absence of Fe, Cu, or other contamination on the hot-chamber diecast surface. Therefore, the next paragraphs discuss the results of the analysis of the die skin microstructure of the hotchamber diecast, and elucidate the correlation between the die skin structure and its corrosion performance.

For the phases in the interdendritic position of a ascast Mg-Al- Zn alloy, Lunder[35] reported that Al-rich alpha phase (surrounding the primary a) plus the beta phase were observed in the interdendritic spacing. A rapidly cooled AZ91 sample (as in the diecasting process) typically consists of primary alpha and an abnormal eutectic of alpha and beta, because of segregation during solidification.[24, 36] Song et al.[24, 37] studied the effect of the microstructure on the corrosion of the cold-chamber diecast AZ91D Mg-Al-Zn alloy, indicating that the phases in the interdendritic spacing were Al-rich alpha (eutectic alpha) and Al^sub 12^Mg^sub 17^ beta A microgalvanic corrosion current is suggested to flow from the beta phase to the alpha phase.[24, 37- 38] The Al-rich alpha phase (eutectic alpha) is considered to be a single phase that has a much higher Al solution concentration than the first-solidified primary alpha,[24, 37,38] despite the fact that no TEM examination of the Al-rich alpha phase was performed. The TEM examinations (as displayed in Figures 9 and 10) yielded various results. A comprehensive characterization of the interdendritic phases by TEM demonstrated that fine Al-Mn or Al-Mn-Mg particles and many fine Al^sub 12^Mg^sub 17^ particles were distributed close to the primary beta phase (Figures 9 and 10). The Mg matrix around primary beta (Figure 10) had a low Al concentration, down to ~4 wt pct. Restated, the so-called Al-rich alpha phase in the diecast chill skin was not a single phase. Rather, it was a composite phase that was composed of mainly fine beta particles, some Al-Mn/Al-Mn- Mg particles, and a low-Al alpha-Mg as matrix. All the intermetalh'c particles are cathodic to alpha-Mg in a corrosive environment.[38- 40] As mentioned in the first paragraph of this section, the corrosion resistance of the diecast Mg alloy was strongly related to the interdendritic phases in die chill skin. Thus, as will be discussed in the following paragraphs, the TEM experiments to elucidate the microstructures of the Al-rich a phase were essential to determining the link between the microstructure (Figures 2, 3, 4, 9, and 10) and corrosion performance (Figures 6 through 8).

A galvanic corrosion current flowed between the Al-enriched alpha phase and the primary Al^sub 12^Mg^sub 17^ beta phase.[36,39] The similar results of Ambat et al.[41] and Aung et al.[42] suggested that beta phase is strongly cathodic to the matrix and can thus act as an effective cathode, causing galvanic corrosion in the anodic alpha matrix. Song et al.[24] schematically represented the microgalvanic corrosion current that flows between primary beta and primary alpha, and also between primary beta and Al-rich alpha, as occurs on the surface of a 6-mm-thick AZ91D plate. Regardless of the TEM results concerning the Al-rich alpha phase that was actually composed of fine beta particles plus an alpha-Mg matrix with low Al concentration (Figures 9 and 10), various studies have suggested that microgalvanic corrosion currents flow only between primary beta and alpha-Mg and between primary beta and the Al-rich alpha phase. In this study, experiments in which the primary beta phase was removed from the surface of the AZ91D diecast plate by Ar+ etching were performed (the results are in Figures 11(a) and (b)). Despite the fact that the die chill layer was free of the primary beta phase, electrochemical tests still indicated that the sample exhibited a high corrosion current density (averaging I^sub corr^ = 1200 [mu]A/cm^sup 2^), which was only slightly better than that (~1600 [mu]A/cm^sup 2^) of the as-diecast sample (Figure 12 for comparison). This work suggests that the primary beta phase on an as- diecast AZ91D chill surface was not the only cathodic phase that dominated the corrosion performance of the material, since many intermetallic particles still existed in the Al-rich alpha. Removing the interdendritic phases (both primary beta and Al-rich a) from a die chill surface significantly increased the corrosion resistance of an as-diecast AZ91 thin plate.

This work elucidated the microstructure of the Al-rich alpha phase, indicating that the Al-rich alpha phase comprised fine beta, Al-Mn particles and the Mg matrix with low Al concentration (Figures 9 and 10). The 8 particles and Al-Mn particles are cathodic to alpha- Mg.[38-40] Based on the electron microscopic observations herein, Figure 13 presents schematic cross sections of the micro structure at the diecast surface and the corrosion caused by the galvanic effect. The primary alpha grains, primary beta, and the Al-rich alpha eutectic phase presented in Figure 13 are greatly magnified. The figure emphasizes the effect of the Al-rich alpha microstructure on corrosion. As schematically displayed in Figure 13(a), the primary beta is not the major cathodic phase that caused the dissolution of the anodic phases due to the microgalvanic corrosion currents that flow between them. Rather, the intermetallic particles in Al-rich alpha participated markedly in the corrosion. As displayed in Figure 13(b), removing the primary beta phase from the diecast surface by ion etching did not reduce the occurrence of galvanic corrosion between phases, since numerous cathodic particles (smaller than 0.5 [mu]m) were present in the Al-rich alpha phase. This can explain that the I^sub corr^ value of the ion-etched sample is similar to that of the as-diecast sample (result in Figure 12). As schematically illustrated in Figure 13(c), the occurrence of galvanic corrosion can be substantially reduced, owing to the removal of the interdendritic phases (both Al-rich alpha and primary beta) by HF/H^sub 2^SO^sub 4^ selected etching treatment. Therefore, the selected etching treatment can markedly improve the corrosion performance of the diecast sample, because the treatments described here removed the highly corrosive phases from the sample's die skin. V. CONCLUSIONS

The following conclusions are drawn from the preceding results and discussion.

1. A hot-chamber diecast AZ91D thin plate with a die chill skin on its surface was severely corroded in chloride solution, whereas a plate with a die skin layer etched in an HF/H^sub 2^SO^sub 4^ aqueous solution to remove interdendritic phases had a substantially lower corrosion rate in 5 wt pct chloride solution. The experimental results on weight loss were consistent with the electrochemical measurements. The GAXRD results of the selected etching sample demonstrated that no X-ray peaks of fluoride compound were obtained in the X-ray spectrum.

2. The die skin layer on the AZ91D diecast thin plate had alpha- Mg dendritic grains with a surrounding interdendritic phase. Based on SEM/EDS results, the interdendritic phase was a nearly composite structure that contained Al-rich alpha-Mg and the primary beta phase. The primary beta on the die skin was irregularly shaped, and did not have a network structure. The TEM examinations revealed that the Al-rich alpha phase in the diecast chill skin was actually not a single phase. Rather, it was a composite phase that was mostly fine beta particles, some Al-Mn/Al-Mn-Mg particles, and a low-Al alpha- Mg matrix.

3. Argon ion etching can be used to remove the primary Al^sub 12^Mg^sub 17^ beta phase alone on the die skin from the surface of the sample, but the alpha-Mg grain and the Al-rich alpha phase were retained. The sample with a die skin layer etched using an ion beam to remove the primary beta phase alone retained a substantially high corrosion rate. The primary beta phase was not the only or the dominant cathodic phase that caused severe corrosion of the as- diecast thin plate. The fine Al-Mn-like particles and fine beta particles that were distributed in the Al-rich alpha phase region might have accounted for the inferior corrosion performance of the sample without primary beta phase on its surface. Therefore, simultaneously removing the interdendritic phases (both primary beta and Al-rich alpha) improved the corrosion performance of the diecast AZ91D thin plate.


The authors thank Trysyntec Technology Cooperation Limited (Taoyuan, Taiwan) for providing the qualified diecast AZ91D panels used in this study. The authors also greatly appreciate Professor S.K. Yen, Department of Materials Science and Engineering, National Chung Hsing University, for providing some very useful discussions. Ms. L. C. Wang, National Sun Yat-Sen University, is acknowledged for her assistance with the TEM works. The authors are also grateful to the National Science Council of Taiwan for financially supporting this work (Contract No. NSC 92-2216-E-005-018).


1. N. Li: Mater. Sci. Forum, 2005, vols. 488-489, pp. 931-35.

2. M. Avedesian and H. Baker: Magnesium and Magnesium Alloys, ASM Specialty Handbook, ASM International, Materials Park, OH, 1999, pp. 1-71.

3. I. Nakatsugawa, H. Takayasu, and K. Saito: in Magnesium Alloys and Their Applications, Int. Congr. of Magnesium Alloys and Their Applications, Munich, Sept. 26-28, 2000, K.U. Kainer, ed., Wiley- Vch. Weinheim, Germany, 2000, pp. 445-50.

4. C.Y. Cho, J.Y. Uan, and H.J. Lin: Mater. Sci. Eng., A, 2005, vol. 402, pp. 193-202.

5. D.A. Jones: Principles and Prevention of Corrosion, 2nd ed., Prentice Hall, Englewood Cliffs, NJ, 1996, pp. 40-74.

6. A. Elbetieha and M.H. Al-Hamood: Toxicology, 1997, vol. 116, pp. 39-47.

7. H. Huo, Y. Li, and F. Wang: Corros. Sci., 2004, vol. 46, pp. 1467-77.

8. A.L. Rudd, C.B. Breslin, and F. Mansfeld: Corros. Sci., 2000, vol. 42, pp. 275-88.

9. K.Z. Chong and T.S. Shih: Mater. Chem. Phys., 2003, vol. 80, pp. 191-200.

10. D, Hawke and D.L. Albright: Met. Finish., 1995, vol. 93, pp. 34-38.

11. C.S. Lin and S.K. Fang: J. Electrochem. Soc., 2005, vol. 152, pp. B54-B59.

12. M.A. Gonzalez-Nunez, C.A. Nunez-Lopez, P. Skeldon, G.E. Thompson, H. Karimzadeh, P. Lyon, and T.E. Wilks: Corros. Sci., 1995, vol. 37, pp. 1763-72.

13. M.A. Gonzalez-Nunez, P. Skeldon, G.E. Thompson H. Karimzadeh: Corrosion, 1999, vol. 55, pp. 1136-43.

14. M. Dabala, K. Brunelli, E. Napolitani, and M. Magrini: Surf. Coal. Technol., 2003, vol. 172, pp. 227-32.

15. M.P. Schriever: Canadian Patent 2,056,159, 1990.

16. J.E. Gray and B. Luan: J. Alloys Compd., 2002, vol. 336, pp. 88-113.

17. G. Hanko, H. Antrekowitsch, and P. Ebner: JOM, 2002, vol. 54, pp. 51-54.

18. J.I. Skar, L.K. Sivertsen, and J.M. Oster: Int. Conf. on Environmental Friendly Pre-Treatment for Aluminum and Other Metals, Sintef, Oslo, Norway, 2004, pp. 1-4.

19. C.E.M. Meskers, A. Kvithyld, M.A. Reuter, and T.A. Engh: Magnesium Technology-TMS Annual Meeting 2006, A.A. Luo, N.R. Neelameggham, R.S. Beals, eds., The Minerals, Metals & Materials Society, Warrendale, PA, pp. 33-38.

20. A. Yamamoto, A. Watanabe, K. Sugahara, S. Fukumoto H. Tsubakino: Mater. Trans., 2001, vol. 42, pp. 1237-42.

21. B.L. Yu and J.Y. Uan: Scripta Mater., 2006, vol. 54, pp. 1253- 57.

22. G. Song, A. Atrens, X. Wu, and B. Zhang: Corros. Sci., 1998, vol. 40, pp. 1769-91.

23. N. Pebere, C. Riera, and F. Dabosi: Electrochim. Acta, 1990, vol. 35, pp. 555-61.

24. G. Song, A. Atrens, and M. Dargusch: Corros. Sci., 1999, vol. 41, pp. 249-73.

25. C. Blawert, E. Morales, V. Heitmann, and W. Dietzel: Light Metals 2005, 44th Annual Conf. of Metallurgists of CIM, Calgary, AB, 2005, J.P. Martin, ed., The Canadian Institute of Mining, Metallurgy and Petroleum, Montreal, Quebec, Canada, pp. 109-26.

26. C. Blawert, V. Heitmann, E. Morales, W. Dietzel, S. Jin E. Ghali: Can. Metall. Q., 2005, vol. 44, pp. 137-46.

27. B.L. Yu and J.Y. Uan: Metall. Mater. Trans. A, 2005, vol. 36A, pp. 2245-52.

28. Reagent Chemicals: American Chemical Society Specifications, 7th ed., American Chemical Society, Washington, DC, 1986, pp. 284- 86.

29. H. Hoche, H. Scheerer, R. Fritsche, A. Thissen, S. Flege, E. Broszeit, C. Berger, H.M. Ortner, and W. Jaegermann: Materialwiss. Werkstofftech., 2002, vol. 33, pp. 132-41.

30. Y.W. Shin, G.A. Sargent, and H. Conrad: Metall. Trans. A, 1987, vol. 18A, pp. 437-50.

31. R.G. Wellman and C. Allen: Wear, 1995, vols. 186-187, pp. 117- 22.

32. ASTM B117: Standard Test Method of Salt Spray Testing, ASTM, Philadelphia, PA, 1990, pp. 20-26.

33. S.J. Pennycook and D.E. Jesson: Ultramicroscopy, 1991, vol. 37, pp. 14-38.

34. H.E. Friedrich and B.L. Mordike: Magnesium Technology: Metallurgy, Design Data, Applications, Springer-Verlag, Berlin, Germany, 2006, pp. 217-68.

35. O. Lunder: Corros. Rev., 1997, vol. 15, pp. 439-69.

36. S. Mathieu, C. Rapin, J. Hazan, and P. Steinmetz: Corros. Sci., 2002. vol. 44, pp. 2737-56.

37. G. Song and A. Atrens: Adv. Eng. Mater., 2003, vol. 5, pp. 837-53.

38. S. Mathieu, C. Rapin, J. Steinmetz, and P. Steinmetz: Corros. Sci., 2003. vol. 45, pp. 2741-55.

39. O. Lunder, J.E. Lein, T.Kr. Aune, and K. Nisancioglu: Corrosion, 1989, vol. 45, pp. 741-48.

40. O. Lunder, M. Videm, and K. Nisancioglu: SAE. SP., 1995, pp. 57-62.

41. R. Ambat, N.N. Aung, and W. Zhou: Corros. Sci., 2000, vol, 42, pp. 1433-55.

42. N.N. Aung and W. Zhou: J. Appl. Electrochem., 2002, vol. 32, pp. 1397-1401.

JUN-YEN UAN, Associate Professor and CHING-FEI LI, Graduate Student are with the Department of Materials Science and Engineering, National Chung Hsing University, Taichung, Taiwan, Republic of China. Contact e-mail: [email protected] BING- LUNG YU, formerly Graduate Student with Department of Materials Science and Engineering, National Chung Hsing University, is now an Engineer with ProMOS Technologies Inc., Taichung, Taiwan, Republic of China.

Manuscript submitted October 29, 2006.

Article published online January 23, 2008

Copyright Minerals, Metals & Materials Society Mar 2008

(c) 2008 Metallurgical and Materials Transactions; A; Physical Metallurgy and Materials Science. Provided by ProQuest Information and Learning. All rights Reserved.