The Effect of Sulfuric Acid Anodization on the Electrochemical Properties of Aluminum Alloy AlSi10Mg Prepared by Selective Laser Melting (2024)

Specimens were fabricated using a SLM*500 industrial 3D printer (SLM Solutions, Germany). The feedstock powder, build parameters, and part cleaning before shipment to MSU have been described in detail in a prior publication. ¹⁵ The specimens were printed from AlSi₁₀Mg source alloy powder with a diameter ranging from 25–65 μm. In total, 10% of the particles were less than 26 μm in diameter, 50% less than 40 μm, and 90% less than 63 μm. According to ASTM B85 (Standard Specification for Aluminum Alloy Die Castings), this alloy powder has the following elemental composition (wt%): Cu (0.6), Mg (0.4–0.6), Fe (≤ 1.3), Sn (≤ 0.10), Ni (≤ 0.5), Zn (≤ 0.5), Mn (≤ 0.35), Si (9.0–10.0) and Al (balance). No independent measurement of the powder was made to determine its elemental composition.

The SLM processing involves melting and fusing metallic powder particles together with a high-power laser beam that is scanned across the build surface. Once a layer of solid metal is created, the tray holding the specimen is lowered and a new powder layer is deposited over the top. This is followed by laser-assisted melting and particle fusion to form the new layer. All specimens were prepared with the following dimensions: 2.54 cm (length) × 2.54 cm (height) × 0.635 cm (thickness), such that the large face (length × height) is the XZ plane orthogonal to the build plane. The XZ plane of all specimens was studied in this work. The SLM system uses 3D optics and four lasers (350 W and 1070 nm) to produce parts with a maximum build space of 500 mm × 250 mm × 365 mm. The inert gas used in the chamber during the specimen builds is Ar to limit oxygen reactivity with the heated aluminum. The following optimized parameters were used for part fabrication: laser power = 350 W, hatch spacing = 130 μm, powder layer thickness = 30 μm, and laser scan velocity = 1650 mm s⁻¹.

Specimen post processing consisted of wire electric discharge machining (EDM), media blasting, and cleaning. No thermal annealing was applied to any of the fabricated specimens. The wire EDM step removes the parts from the build plate. The media blast step helps remove semi-sintered particles from the specimen, reduces the surface roughness some, and provides a matte finish to the exterior. The alloy specimens were blasted with sand at a pressure from 50–70 psi to remove loose powder particles. Cleaning was then performed consisting of the five general steps: (i) high pressure spray with detergent and rinse with water, (ii) ultrasonication in detergent followed by a water rinse, (iii) ultrasonication in detergent followed by a rinse and drying with a stream of N₂ gas, (iv) vacuum oven drying and (v) packaging the cooled specimens into a nylon bag for transport to MSU. This constituted the "as-received" specimens.

Pretreated specimens were potentiostatically anodized in 9.8 vol% H₂SO₄. A DC power supply (Tenma) was used to control the applied voltage for anodization, as described previously. ³³ The positive terminal was connected to the AlSi₁₀Mg specimen (anode) with a wire and alligator clip. The cathode was a piece of SS304L plate with dimensions of 2.5 cm × 1.7 cm and a 0.1 cm thickness. The electrodes were placed parallel to one another at a fixed distance of 5.7 cm using a clamp secured to a ring stand. Anodization was performed in a 400 mL beaker filled with 9.8 vol% H₂SO₄. The electrodes were lowered into the bath such that approximately 1.9 cm (about ¾ of the sample) of each was immersed. The anodizing process began at 0 V ramping to 15 V at a rate of 5 V min⁻¹ in 1 V steps. After this 3-min ramp, the voltage was held at 15 V for 20 min giving a total anodization time of 23 min. These parameters were selected based on prior work anodizing wrought AA2024-T3 to produce an oxide weight of ∼1000 mg ft⁻². ³³ The anodized specimen was removed from the beaker and rinsed with ultrapure water. To fully remove slowly dissolving anodization smut, the specimen was immersed in ultrapure water for at least 20 min with periodic agitation. Finally, the specimen was then dried with a stream of N₂ and stored in a covered petri dish in the lab atmosphere for at least 24 h before further testing.

The oxide weight after anodization was determined by an acid stripping method, as described in ASTM B137 (Standard Test Method for Measurement of Coating Mass Per Unit Area on Anodically Coated Aluminum). The specimen was weighed using an analytical balance (±10 μg) before and after stripping in a solution of chromic acid anhydride (20 g l⁻¹) and orthophosphoric acid (35 mL l⁻¹) that was heated to 100 °C. The anodized specimen was immersed in the stripping solution for 5 min, rinsed in ultrapure water, dried with N₂, and weighed. The acid dissolution was repeated until the mass loss was negligible (< 0.010 mg). The weight loss (mg) is attributed to the dissolution of Al₂O₃ oxide layer. The oxide weight per unit area is reported in mg/ft², as per ASTM B137.

Rapid melting and particle fusion in the SLM process leads to a nonequilibrium microstructure. Studies of the SLM AlSi₁₀Mg microstructure have been reported in the literature. ^{5,6,9,12,34,35} The SEM micrograph in Fig. 1 presents a representative example of the unanodized alloy microstructure. The microstructure is characterized by a cellular α-Al phase that is surrounded by a more fibrous Al-Si eutectic phase. ^5,6,8–13 Within the melt pools, the α-Al cells all have an aspect ratio greater than 1.0 with the long dimension of the cells in the MP Coarse region being ca. 1–3 μm and the long dimension in the MP Fine region being generally 1 μm, or less. The dendritic morphology is comprised of cells with a columnar orientation in the build direction due to the unidirectional heat transfer during laser melting and solidification. ^5,6 The Al grain size is smaller in the center of the melt pool (MP fine) and larger at the edges (MP coarse) because of the large thermal gradients and different cooling rates. As each additional powder layer is fused to the previous layer, the Si network at the surface of the build is recrystallized in the heat affected zone (HAZ). Note that the HAZ region consists of a disrupted or broken Al-Si eutectic phase surrounding the α-Al cells. This is an area reported to have higher galvanic activity and a greater vulnerability to corrosion initiation. ¹⁰ The grain structure and micro texture of the SLM AlSi₁₀Mg alloys resemble materials used other reports for this SLM alloy, but are distinct from reported data for the cast alloy. ^5,6,34 The reason for the smaller, columnar grains that develop in the build direction is the rapid and directional solidification that occurs in the SLM process. ^5,6,34 Electrochemically, the interface between the α-Al matrix and the continuous or broken Al-Si eutectic phase constitutes a region where the passivating anodic coating will not form completely leading to sites were solution penetration can occur and crevice or localized pitting corrosion can initiate.

Three surface pretreated AlSi₁₀Mg alloy specimens were anodized with the current measured at different time points during the potentiostatic process. The current density was calculated by dividing the measured current by the total geometric area of the specimen immersed in solution including the edges and rear face. Sampled anodization current density vs time curves for three different specimens are shown in Fig. 2. The current density increases proportionally with the voltage during the ramp phase. The maximum current density is achieved at the beginning of the hold phase at 15 V of between 12 and 15 mA cm⁻². The 15 V used herein is a typical anodization voltage for aluminum and was used in our prior work with anodized AA2024-T3. ³³ The current then declines over the next 3 min, trending toward a more constant value of 6–7 mA cm⁻² for the duration of the anodization. During the ramp, the proportional increase in current density is because of the higher rate of oxidation with increasing voltage. Once the hold begins at 3 min, the current decreases because of the maturing surface coverage and increasing thickness of the oxide. Finally, at times beyond 10 min, the near steady-state current of 6–7 mA cm⁻² reflects a constant rate of oxide growth. For comparison, the anodization current densities for wrought AA2024-T3 alloys under the same conditions are 30–60 mA cm⁻², as reported in prior work, reflective of the greater surface area of aluminum exposed in the wrought alloys. The Al-Si eutectic phase reduces the effective surface area of the SLM AlSi₁₀Mg specimens exposed to the anodizing bath per specimen geometric area.

The electrochemical formation of porous aluminum oxide is shown in this redox reaction:

The equation indicates 1 mol of oxide is produced for every 6 mol of electrons passed. The theoretical mass of oxide (m_ox ) formed can be estimated from the total charge (Q) passed using Faraday's law in Eq. 2

In this equation, M_ox is the molar mass of Al₂O₃ (102 g mol⁻¹), n⁻¹ is the number of electrons to produce one mole of oxide (6), and F is the Faraday constant (96,500 C mol⁻¹). To account for other electrochemical reactions, the Faradaic efficiency (η) is included in the calculation. A value of η = 0.6 is commonly assumed, based on previous work by Veys-Renaux ¹⁹ and Wu. ²¹ Dividing the oxide mass by the anodized surface area yields an oxide weight of 918 ± 42 mg ft⁻² (mean ± standard error of the mean) for multiple specimens (n = 3). The oxide weight, as measured by the chemical stripping method, was a little lower at 698 ± 29 mg ft⁻² or 7.5 ± 0.3 g m⁻². The discrepancy between the two values suggests a lower Faradaic efficiency than the assumed η = 0.6 value. If the oxide weight from ASTM B137 is used in Eq. 2, the calculated Faradaic efficiency is η = 0.46 ± 0.03. The lower η can be explained by some charge being passed for the oxidation of Si to SiO₂, which has been reported by Revilla and coworkers. ²⁰ To summarize their work, SLM and the diecast AlSi₁₀Mg alloys were anodized galvanostatically in 9.8% H₂SO₄ followed by characterization using XPS depth profiling. For the SLM prepared alloy, the XPS profiling revealed that SiO₂ was present throughout the oxide. In contrast, the die cast alloy contained no subsurface Si oxide. The presence of the Al-Si eutectic phase in the SLM prepared alloy causes the reduced Faradaic efficiency. ²⁰

SEM micrographs of an anodized specimen taken in the gentle beam (GB) mode are presented in Fig. 3. In Fig. 3a, the anodized surface is viewed at the same magnification as in Fig. 1 for comparison. The Al-Si eutectic phase network and columnar α-Al grain structure are evident inthe fine and coarse melt pool regions and the HAZ region in between the sintered layers (yellow dashed lines). The α-Al cells are, however, larger in the MP Course region. A higher magnification micrograph of the red boxed area is shown in Fig. 3b. On the surface, the Al-Si eutectic phase network remains.

Intact within the coarse melt pool region. Closer investigation of the Si and Al regions in Fig. 3c reveals a porous oxide has grown across and into the α-Al grains that are surrounded by a more discontinuous Al-Si eutectic phase. A uniform oxide pore diameter of ∼20 nm is seen. In some places, the Si phase remains intact, but in other places it looks to be oxidatively damaged. This damage is apparent when comparing the anodized surface in Fig. 3c with the unanodized surface shown in Fig. 3d. The Al phase in Fig. 3d is nonporous and is lined by mostly a continuous and nodular shaped Si eutectic phase. In some areas where the anodized α-Al connects with the Si eutectic phase, there are larger depressions and voids in the aluminum that extend into the alloy and below the Si phase. These can be seen in the left center of the micrograph presented in Fig. 3c. This is a notable feature, as defects along the α-Al and Si eutectic phase interface are locations for electrolyte solution penetration and corrosion initiation.

A SEM micrograph of a cross section of an anodized AlSi₁₀Mg specimen is shown in Fig. 4. The oxide thickness is indicated by the yellow bars with values of 4.5 to 6.0 μm over the region probed. The mean ± standard error of the oxide layer thickness is 5.4 ± 0.3 μm (average ± standard error of the mean). Although there is some variation in the oxide thickness, the oxide layer appears continuous across the alloy surface. This indicates that there is good physical and chemical contact between the oxide and the underlying aluminum.

The micrograph in Fig. 5a shows that large sections of the Al-Si eutectic phase remain intact through the growing oxide. The brighter appearing features running diagonally through the oxide cross section are the Al-Si eutectic phase. Near the top of the micrograph, the resin/oxide boundary represents the outer surface of the anodic coating. An isolated alloy defect that has been entrapped within the oxide layer is shown in greater detail in Fig. 5b, labeled herein as a microvoid. The micrograph in Fig. 5b shows the yellow boxed region in Fig. 5a labeled b. This bulk alloy defect likely resulted from the entrapment of moisture or gas during the SLM fabrication process. Weingarten et al. reported these internal porosity artifacts as hydrogen porosity. ³⁶ These are sites of interest in the AM materials. ³⁷ This fabrication defect was not commonly observed in the AlSi₁₀Mg specimens used herein. One microvoid was found in the oxide layer (Fig. 5b), and another in the alloy (Fig. 5d), for comparison. This gives insight into how the oxide layer formation occurs around fabrication defects. Along the top edge of the microvoid, the aluminum oxide pores branch out horizontally, then turn downward (yellow arrows). This type of oxide coating morphology has been noted by Revilla et al. in a study of porosity artifacts in the SLM alloy. ²⁹ The pores grow perpendicular to the walls of the void because the porous oxide layer allows for electrolyte solution penetration into the micro void, filling it. Once the electrolyte contacts the unanodized bottom and sides of the void, the oxide layer grows perpendicular to the interface. Finally, as the entire surface of the void becomes oxidized, the anodizing front (oxide/metal interface) passes below the microvoid, and anodizing continues downward into the alloy. The oxide/metal interface is shown in greater detail in Fig. 5c, which is the yellow boxed area in Fig. 5a labeled c. The yellow arrows show the orientation of the pores are not perfectly straight as would be seen for a cast alloy (e.g. Revilla et al.). ²⁰ In contrast, the observed oxide/metal interface is not uniform in depth, but rather jagged at the high magnification. This can be explained by the tortuous path that the aluminum oxide pores must follow when forming through the Si-rich network. Key parameters of oxide coatings for corrosion protection are the thickness, pore structure, pore size distribution, and adhesion to the aluminum base. Increased thickness and decreased porosity, either by sealing or adjustments in the anodization parameters, enhance the corrosion protective properties.

The backscattered electron micrograph in Fig. 6a was analyzed using EDXS mapping. Elemental maps for O, Al and Si are displayed in Figs. 6b–6d. The maps provide qualitative insight on the location of different elements in the oxide cross section. The X-ray analysis data was only used qualitatively. The O Kα1 intensity dominates the top half of the map where the oxide coating resides. Based on the scale bar and the presence of O, the oxide layer is approximately 5 μm thick. The Al Kα1 signal is present in both the top and bottom halves of Fig. 6c. The signal for Al is lower in the oxide than in the bulk alloy as expected given the lower density of the element. Likewise, the Si Kα1 intensity is present throughout the fully mapped area, and aligns with eutectic network seen in Fig. 6a. EDXS data (not presented here) also revealed the presence of S, from SO₄ ²⁻, incorporated within the oxide. The element was present in the outer oxide, but the spatial resolution was insufficient to confirm the presence near or within the barrier layer. It has been reported that the sulfate content is low at the initial stage of anodization in H₂SO₄ and increases with anodization time to a near constant value. ³⁸ This is explained by the greater inward mobility of O₂ ⁻ ions relative to SO₄ ⁻² ions under the electric field during anodizing. From the microscopy data at hand, the oxide coating is about 5 μm thick for these anodization conditions, but the thickness of the all-important barrier layer at the base of the oxide coating cannot be determined.

Mg is added to the alloy for strengthening. It largely diffuses into the Al matrix in these SLM AlSi₁₀Mg alloys. Some Mg₂Si secondary phases can form at the Al-Si eutectic phase boundary depending on the processing conditions. ^5,39 Our SEM and EDAX analysis of the specimens used in this work did not reveal any of these secondary phases.

Various measurements were made to investigate the effect of the oxide layer on the electrochemical properties of the AlSi₁₀Mg alloy. The open circuit potential (OCP) vs time was recorded and example curves are presented in Fig. 7a. The OCP for the anodized specimen drifted positive over the first 1000 s before reaching a maximum and then stabilizing at a less positive value of ca. −0.450 V. The OCP for the unanodized specimen started at a more active potential and did not change much over time stabilizing at ca. −580 V after 2000 s. Table I presents a summary of the stabilized OCP values for the two specimen types. The nominal value for the anodized specimens is more noble than the value for the unanodized specimens by about 150 mV. This trend is consistent the passivating nature of the anodic coating.

Anodic and cathodic potentiodynamic polarization curves for multiple anodized (black curves) and unanodized (red curves) specimens are presented in Fig. 7b. It is clear from the data that both anodic and cathodic currents are suppressed by the anodic oxide coating. For example, the anodic current at −0.1 V is suppressed by 15× after anodization while the cathodic current at −0.8 V is suppressed by 215× (see Table I). These potentials were selected arbitrarily for reporting current differences. The data indicate that the oxide coating slows the redox reaction rates for both aluminum oxidation (Eq. 3) and dissolved oxygen reduction (Eq. 4).

In the anodic polarization curves for the anodized specimens, there is no onset of localized oxide breakdown and stable pit formation and growth (i.e., sharp anodic current increase with potential) out to 0.5 V for any of the specimens, consistent with the oxide coating passivating the underlying surface. Beyond −0.1 V, the specimens exhibit a passivation current density, j_pass, of ca. 10 μA cm⁻². The passivation current density is about 15× lower after anodization (see Table I). While no breakdown was observed on any of the anodized specimens, two of the three unanodized specimens showed signs of native oxide layer breakdown. At high anodic potentials, the chloride-containing electrolyte causes the breakdown of the native oxide layer on aluminum. ⁴⁰ The oxide breakdown occurs locally, and the active Al oxidizes rapidly forming stable growing pits. The pitting potential (E_pit) marks the point at which the current increases significantly above the passivation current reflecting stable pit formation and growth. One unanodized specimen exhibited no evidence for oxide coating breakdown out to 0.5 V vs Ag/AgCl. Two unanodized specimens, however, exhibited breakdown or E_pit at 72 and 270 mV vs Ag/AgCl, respectively. This variance from specimen to specimen is attributed to different levels of imperfections and defects in the native oxide layer that forms on the alloy surfaces.

The oxygen reduction reaction (ORR) is assumed to be the primary cathodic reaction in naturally aerated electrolyte at potentials from the OCP down to −1.1 V vs Ag/AgCl. The anticipated oxygen reduction reaction is shown in Eq. 4.

Evidence that the current at these potentials is due to the reduction of dissolved oxygen is the attenuation in the current that was observed underdeaerated conditions (data not shown). The cathodic scan begins at the OCP and extends down to −1.2 V vs Ag/AgCl. The current density near the OCP is two orders of magnitude lower for the anodized specimens, as compared to the unanodized control. Table I reveals the nominal cathodic current density at −0.8 V is 215× lower for the anodized specimens. The current for both specimens, but particularly for the anodized ones, increases negative of −1.1 V vs Ag/AgCl due to the onset of reduction of Al₂O₃ and the hydrogen evolution reaction (HER).

Table II summarizes linear polarization resistance (R_p) values for anodized and unanodized alloy specimens. R_p increases nominally to 3.07 (± 0.43) ×10⁶ Ω·cm² after anodization from 9.99 (± 2.69) ×10³ Ω·cm². This increase of 279× reflects the increased alloy corrosion resistance because of the passivating anodic oxide coating. Based on the polarization curve currents, the most significant effect of the oxide layer is on suppression of the oxygen reduction reaction current, as seen in in Fig. 7.

Tafel analysis was performed on the potentiodynamic polarization curve data to determine how the Tafel slopes, β (mV/decade of current, reaction mechanism), and corrosion current density, j_corr (A/cm², measure of the corrosion rate) are affected by the oxide coating. Tafel plot extrapolation was used to extract the j_corr and the corrosion potential, E_corr, values. No iR correction was applied to the η–log j data. The assumption is that a single cathodic reaction is occurring, namely the reduction of dissolved oxygen.

Table III presents a summary of the Tafel analysis data for anodized and unanodized specimens (n = 3 of each). Nominally, the Tafel slope is slightly lower for the anodic curves and a slightly higher for the cathodic curves after anodization. Overall, though, the anodization does not significantly alter the Tafel slope for either reaction. Therefore, the mechanisms for the two reactions are not changed by the oxide coating. However, the oxide coating does significantly decrease the currents for both reactions with the nominal j_corr value being 133× lower than the nominal value for the unanodized control specimens. This effectively translates into a two order of magnitude decrease in the corrosion rate at E_corr. This decrease in j_corr is consistent with the average suppression of the anodic and cathodic potentiodynamic polarization curves (Table I) of 115× and the nominal increase in the R_p value of 279×.

Electrochemical impedance spectroscopy measurements were performed at the OCP in naturally aerated 0.5 M Na₂SO₄ + 0.01 M NaCl using anodized (black) and unanodized (red) AlSi₁₀Mg specimens. Figure 9 presents impedance spectra in the forms of Bode diagrams of (a) total impedance and (b) phase shift vs log frequency. A high degree of reproducibility between the replicate measurements is seen for each specimen type. At all frequencies, except the highest, the impedance of the anodized specimens is greater than the unanodized ones. At the highest frequencies, the Bode diagrams for both specimen types exhibit a constant impedance of ca. 20–50 Ω•cm² with a phase angle approaching 0°. This corresponds to the series resistance, which is the sum of the ohmic resistances of the metal alloy and electrolyte solution. This resistance is similar for both specimen types, as expected. At middle frequencies, the impedance increases with decreasing frequency while the phase shift approaches −70° for the anodized and −80° for the unanodized specimens. The phase angle for the anodized specimens is −60° to −80° over the frequency range from 10⁴ down to 10⁻² Hz while the unanodized specimens exhibit a phase shift maximum at about 50 Hz. The larger phase angle for the anodized specimens and the fact that the phase angle remains near −80° to lower frequencies is consistent with the formation of thick oxide coating with capacitive properties. ^19,23,33

At the lowest frequencies for the anodized specimens, there is a significant increase in the low frequency impedance at Z_{0.01 Hz} with a value of ca. 10⁶ Ω·cm². A value of 10⁹ Ω·cm² would be reflective of a very high level corrosion protection. ⁴¹ The 10⁶ Ω·cm² value reflects a good level of corrosion protection provided by the anodic coating that could be further improved by sealing. ^22–25,33 Furthermore, the 10⁶ Ω·cm² value is 43× larger than the values of 10–30 kΩ·cm² observed for the unanodized control specimens. The increased Z_{0.01 Hz} values are attributed to the suppression of Faradaic reactions by the oxide coating.