Effects of Deformation Texture and Grain Size on Corrosion Behavior of Mg-3Al-1Zn Alloy Sheets.

Effects of deformation texture and grain size on mechanical properties and corrosion behavior of the Mg-3Al-1Zn (AZ31) alloys were systematically investigated. The results revealed that the ultimate tensile strength (UTS) and fracture elongation (FE) significantly increased from 232 to 273 MPa and 12.5 to 26.4%, respectively. According to the immersion and electrochemical measurements, the results indicated that the corrosion resistance of the alloy was improved obviously. Via electron backscattered diffraction (EBSD), corrosion morphology showed that the propagation of local pitting corrosion was suppressed, which is ascribed to grain refinement, higher texture intensity, and lower dislocation density after rolling and subsequent annealing.

Introduction

In recent years, magnesium (Mg) alloys are considered as the lightest structural materials that can provide tremendous weight-saving potential in various fields, such as aircraft, high-speed rail, and automotive industries.[1,2] Also, Mg–3Al–1Zn (AZ31) alloy has become one of the most popular commercial Mg alloys due to their excellent mechanical properties, machine-ability, and cast-ability.[3,4] However, Mg-based alloys usually suffer from poor corrosion resistance, which is one of the important obstacles that limit their widespread applications. This is mainly due to the high reactivity of the magnesium alloy and the unstable passivating characteristic film formed on the alloy surface upon atmospheric exposure.[5,6] Therefore, it is great significance to study the factors that affect the corrosion characteristics and corrosion mechanism of magnesium alloys. On the one hand, alloying can be used to enhance the corrosion resistance of magnesium alloys.[7] For example, Pan et al. reported that Mg–Al–Mn alloys have the best corrosion resistance when 2 wt % of Ca was added; it is mainly because the discontinuous (Mg,Al)2Ca phase could act as an anode to protect the Mg substrates.

On the other hand, deformation can also enhance the corrosion resistance of magnesium alloys. Hot-rolling is a traditional deformation processing that can not only improve mechanical properties but also enhance corrosion resistance of various metal materials.[8−11] For example, Cheng et al.[8] reported that the corrosion resistance of the Mg–6Bi–2Sn alloy is obviously improved after rolling, mainly attributed to the internal microstructure of the alloy, grain refinement, advantageous crystal orientation, and finely distributed second-phase particles. Moreover, the corrosion resistance of the Mg–5Li–1Al alloy has been enhanced by hot-rolling, which was ascribed to the evolution of texture, the appearance of twins, and protective oxide surface films.[9] In general, the strong basal texture and the lower surface energy of the Mg alloy can result in the increase of corrosion resistance.[10,11] Hence, combined heat treatment and hot-rolling can significantly improve the properties of magnesium alloys. After rolling and subsequent annealing, the AZ31 sheet exhibits high strength and good ductility, mainly owing to the homogenous microstructure with weak basal texture and fine grains.[12,13]

Moreover, heat treatment also can make the precipitation phase along the grain boundaries dissolve into the Mg matrix, which could decrease the formation of the microcathode.[3] Heat treatment can be carried out before or after extrusion. On the one hand, corrosion resistance was enhanced due to the decrease of the second phase size and more uniform distribution by the solution treatment before extrusion.[14] On the other hand, solution treatment of the extruded magnesium alloy sheet can decrease the corrosion rate.[15] Meanwhile, annealing treatment can decrease the dislocation density and twins produced in the matrix by the severe plastic deformation, resulting in further enhancement of the corrosion resistance.[16,17] However, Ha et al.[18] demonstrated that the rate of H2 evolution was accelerated because of the increased area fraction of grain boundaries. Furthermore, such microstructural evolution in Mg alloys could have a significant influence on their corrosion behavior after rolling and subsequent annealing.

So far, it has not been clearly understood the effect of grain boundaries during the process of Mg alloy corrosion, as well as the dominant role of grain refinement and texture strength in the increase of corrosion resistance. In this paper, we obtained a homogenous microstructure with different grain sizes and deformation textures by combining thermomechanical deformation and annealing treatment on the AZ31 alloy. Moreover, the microstructural evolution, mechanical properties, and corrosion behavior of the samples were further characterized. Besides, the morphology after removing corrosion products was analyzed by electron backscattered diffraction (EBSD). The results show that the initiation of corrosion mainly occurs inside of the grains and the propagation of local pitting corrosion was suppressed by grain boundaries.

Experimental Methods

Material Preparation

The as-extruded AZ31 magnesium alloy sheets were starting materials in this study. Mg–3Al–1Zn-based alloys were fabricated from commercial pure Mg (99.85%), Al (99.90%), and Zn (99.90%). First, alloying components were melted at 700 °C completely under the protection of a mixed gas atmosphere in an electric resistance furnace containing CO2 and SF6 and then poured into a preheated cylindrical steel mold with a dimension of ϕ 94 × 20 mm2. The cast rods were machined to a diameter of 90 mm followed by subsequent homogenization treatment at 430 °C for 5 h in a resistance furnace. The rods were subjected to an extrusion process at 390 °C with an extrusion ratio of 33.[19] Eventually, as-extruded slabs were formed with a cross-section size of 40 mm (width) × 4.8 mm (thickness). Chemical composition of the as-extruded alloys (Al of 2.56 wt %, Zn of 0.84 wt %, with the balance being Mg) was analyzed by an optical spectrum analyzer (ARL 4460, Switzerland). Solution treatment was carried out on the as-extruded alloy slabs at 415 °C for 20 h, and then, they were cut into plates by wire electrical discharge machining with dimensions of 40 mm (length) × 15 mm (width) × 4.8 mm (thickness). Subsequently, traditional hot-rolling was performed on the as-received AZ31 alloy sheets at 350 °C and repeated for seven passes with a reduction of ∼20% per pass. The thickness of the final rolled plate was 1 mm, and total rolling reduction of ∼79% was achieved without edge cracking. Then, the as-rolled plates were annealed-treated for 30 min at 400 and 200 °C, and the samples are referred to as AT 400 and AT 200, respectively. Schematic illustration of the rolling and annealing process is presented in Figure .

Figure 1

Schematic illustration showing the process of rolling and annealing.

Sample Characterization

Microstructures of AZ31 alloys in solution treatment, as-rolled, and annealed conditions were observed by a scanning electron microscope (SEM; ZEISS EVO18, Germany). The SEM characterization of metallographic samples was performed by pregrinding with 2000-mesh SiC papers, polishing with 0.5 μm diamond pastes, followed by chemical etching in acetic picric solution (5 g of picric acid, 5 mL of acetic acid, 10 mL of distilled water, 80 mL of ethanol) for about 20 s. Grain orientation and microtexture evolution were investigated by a SEM (VEGA 3 XMU, TESCAN, Czech), which is equipped with an Oxford Instruments NordlysNano electron backscattered diffraction (EBSD) detector. The data was collected and analyzed by AZtec and Channel 5.0 software. Moreover, EBSD characterization was performed at 20 kV, with a sample tilt angle of 70°, 0.8–0.9 μm scan steps, and scan area of 250 × 250 μm2. EBSD samples were prepared by electropolishing special polishing solutions (AC2) at 20 V for 90 s under −30 °C.

Tensile samples of dimensions 30 mm (length) × 10 mm (width) × 1.0 mm (thickness) were cut off from as-received AZ31, as-rolled, and annealed alloy plates parallel to the rolling direction (RD) and extrusion direction (ED). Tensile tests were performed at room temperature (RT) under a strain rate of 1.0 × 10–3 m s–1 by a material testing system (MTS; 810 testing machine). Each sample was tested at least 3–5 times so as to verify the good repeatability of stress–strain curves. The Vickers hardness experiment was carried out with a load of 200 g (Hv0.2) by using a microhardness tester (1600–5122 VD), and the duration time is 15 s at room temperature. The hardness value of each sample was measured ten times at different locations, and the test surface is perpendicular to the rolling direction.

Immersion and Electrochemical Measurements

The corrosion behavior of each sample was evaluated through an immersion test and electrochemical experiment. The 3.5 wt % NaCl solution (mL) and the sample surface area (cm2) were 100/1, more information of which can be found in the literature.[17] Samples were suspended in the middle of a funnel and placed into a beaker containing sodium chloride, and evolved hydrogen was collected by a burette at 25 °C for 7 days, which also measured the corrosion rate of the sample simultaneously by calculating the weight loss. Corrosion products were removed by immersing the sample in a mixed cleaning solution of chromic acid for 5 min, containing 200 g/L CrO3 and 2 g/L AgNO3. Then, the samples were cleaned with ethanol and dried with a blower. The dried samples before immersion and after removing corrosion products were weighted. The corrosion rate, CW (mm y–1), was evaluated by the following equation[20]where Δm (mg) is the sample weight loss before the immersion test and removing corrosion products, A is the exposed area of the samples surface, and t (d) is the immersion duration.

Electrochemical polarization tests were conducted at 25 °C in 3.5 wt % NaCl solution using an electrochemical workstation (Gamry reference 600, America). The experiment was performed at a typical three-electrode cell, consisting of a working electrode (samples) with an exposed surface area of 1 cm2, a saturated calomel electrode (SCE) as the reference electrode, and a platinum plate (1.0 × 1.0 × 0.1 mm3) as the counter electrode. Open-circuit potentials (OCPs) of samples were recorded in the NaCl solution, and the time duration was maintained at 30 min to obtain a stable state before the test of electrochemical impedance spectroscopy (EIS). The equivalent circuit model was used to analyze the EIS data by ZSimpWin software. The sinusoidal signal had an alternating current (AC) amplitude of 5 mV, and the frequency range is 105–10–2 Hz. Potentiodynamic polarization curves were gathered from −500 to 500 mV relative to OCP with a sweep rate of 0.5 mV s–1. Corrosion potential (Ecorr, V) and corrosion current density (icorr, A cm–2) were calculated by Tafel extrapolation according to the polarization curve. All immersion and electrochemical tests were carried out at least 3 times, and all curves with a good repeatability are presented.

Results and Discussion

Microstructural Characterization

Figure shows the microstructure images of SEM and the corresponding grain size (GS) distribution of as-received AZ31, as-rolled, AT 400, and AT 200 alloy sheets. In Figure a, only few Mg17Al12 particles distributed in the as-extruded AZ31 alloys after solution treatment. While rolling seven passes with a total reduction of ∼79%, the grain size of alloy sheets was reduced to ∼8 μm due to dynamic recrystallization (DRX), and some deformation twins were created. Moreover, fewer second phases (Mg17Al12) in the samples were found after annealing treatment, and the samples present an equiaxed recrystallization microstructure, as shown in Figure c,d. As shown in Figure e–h, a relatively large average grain size of solution-treated samples of ∼22 μm was found; the AT 400 and AT 200 samples’ average grain sizes were ∼11 and ∼4 μm, respectively. The grain size distributions are statistical analysis (average linear intercept measured with SEM microscopy). As can be seen from aforementioned results, the grains size refined obviously after rolling and then annealing.

Figure 2

SEM micrographs of (a) as-extruded AZ31 plate after solution treatment, (b) as-rolled, (c) AT 400, (d) AT 200, and the corresponding grain size distribution (e–h).

In addition, the grain orientation and microtexture evolution of the samples are characterized in Figure via EBSD. As shown in Figure a,c,d, most grains oriented with c-axes in ⟨001⟩ were observed in the solution-treated and annealed samples. In addition, the grain size is in good accordance with the aforementioned SEM images and the corresponding grain size distribution. Moreover, their corresponding {0002} pole figures exhibited a strong basal texture, as shown in Figure e,g,h. The EBSD orientation maps of the as-rolled sample with a total reduction of ∼79% are shown in Figure b. The unrecognized area (black area) is due to the large number of dislocations inside of the rolled sample. As shown in Figure f, the texture intensity of as-rolled alloy sheets went up from 17.1 to 34.7. The difference of temperature in the process of solution treatment and annealing can cause recrystallization, resulting in a basal texture weakness.[21]

Figure 3

EBSD orientation maps of (a) as-received AZ31, (b) as-rolled, (c) AT 400, and (d) AT 200 alloy sheets and the corresponding {0002} pole figurers (e–h). Low- (2° < θ < 15°) and high- (15° < θ < 90°) angle grain boundaries are depicted by gray and black lines in EBSD orientation maps, respectively.

To further investigate the microstructure, kernel average misorientation (KAM) maps of the as-received AZ31, as-rolled, and annealed samples are shown in Figure . The corresponding number fraction of KAM is shown in Figure S1. KAM values in terms of the density of geometrically necessary dislocations reveal the local misorientation level.[22] The results show that the as-extruded AZ31 plate after solution treatment leads to a lower KAM value as shown in Figure a. However, a high density of dislocations (green and black area) was generated after seven passes rolling, and the results are displayed in Figure b. The microstructures of AT 400 and AT 200 samples consist of fully recrystallized grains, and due to static recrystallization (SRX), the dislocation density of alloy sheets decreased significantly as shown in Figure c,d. As mentioned above, the main microstructural evolutions caused by hot-rolling and annealing are grain refinement, favorable crystal orientation, and low density of crystalline defects.

Figure 4

KAM maps of samples in different conditions: (a) as-received AZ31, (b) as-rolled, (c) AT 400, and (d) AT 200 alloy sheets.

Effects of Microtexture Evolution on Tensile Properties

The engineering stress–strain curves and the results of Vickers hardness (Hv) measurement are presented in Figure . The values of ultimate tensile strength (UTS), yield strength (YS), fracture elongation (FE), and Vickers harness (Hv) are listed in Table . Obviously, compared with the as-received AZ31 sample, the UTS and YS of the sample after seven passes rolling increased from 232 to 313 and 124 to 223 MPa. First, this is mainly ascribed to the presence of high-density dislocation in the as-rolled alloy sheets as shown in Figure b. On the other hand, after hot-rolling, the sheet shows a strong basal texture and the basal plane is parallel to the tensile axis, which dramatically increased the critical resolved shear stress (CRSS).[13,23] Therefore, the FE values drastically decreased from 12.5 to 6.2%. Mg and its alloy’s lack of slip systems resulted in a large Taylor factor, and hence the stress of Mg alloy sheets exhibits a strong dependence on grain size.[13] In the present study, as for the AT 200 alloy sheets, the average grain size reached ∼4 μm, which is much lower than that of the as-received AZ31 of ∼22 μm. The increase of UTS and YS is mainly ascribed to grain refinement strengthening resulting from hot-rolling and subsequent annealing. According to the Halle–Petch relationship (σs = σ0 + kd–0.5), the decrease of grain size results in more grain boundaries that can improve the impediment of dislocation movements. It should be noted that the AT 200 alloy exhibited an FE of 26.4%, which is 2.1 times higher than that of the as-received AZ31 alloy. Furthermore, the microstructures of AT 400 and AT 200 were refined, and the Vickers hardness values of AZ31 alloys increased from 58.5 to 63.5 and 68.1 Hv, respectively.

Figure 5

Engineering stress–strain curves (a) and (b) Vickers hardness (Hv) measurement results of the solution-treated AZ31, as-rolled, AT 400, and AT 200 alloy sheets.

Table 1

Comparison of the Room-Temperature Properties of Solution-Treated AZ31 Sheet with Different Processing

sample	GS (μm)	UTS, σb (MPa)	YS, σs (MPa)	FE, δf (%)	VH (Hv)
AZ31	∼22	232–5+7	124–3+4	12.5–0.4+0.6	58.5–1.2+1.8
as-rolled	∼8	313–5+4	223–4+5	6.2–0.3+0.4	93.7–1.7+2.1
AT 400	∼11	263–2+3	198–2+3	25.1–0.5+0.3	63.5–2.5+2.3
AT 200	∼4	273–1+2	205–1+2	26.4–0.4+0.3	68.1–1.4+0.9

Hydrogen Evolution and Mass Loss

The hydrogen evolution volume and corrosion rate (CW) of all of the samples are illustrated in Figure . As shown in Figure a, the collected hydrogen volume of each sample presents a nonlinear relationship with immersion time. During the early period, the rate of H2 evolution is faster. However, with the increase of reaction time, the H2 evolution rate slows down. This phenomenon is mainly attributed to the increased pH of the solution and the protection of corrosion products. The hydrogen evolution volume of the AT 200 sample was lower than that of the as-received AZ31, as-rolled, and AT 400 samples throughout the immersion period. As shown in Figure b, the weight loss measurement was used to calculate the daily corrosion rates (CW) of each sample by using eq . Moreover, the correlated average corrosion rates are listed in Table . The values of average corrosion rates are 2.35, 1.51, 1.23, and 1.05 mm y–1 for the as-received AZ31, as-rolled, AT 400, and AT 200 alloy sheets, respectively. The results clearly indicate that the corrosion resistance of the as-received AZ31 alloy was improved by combining hot-rolling with annealing.

Figure 6

(a) Hydrogen evolution volume and (b) corrosion rates (CW) of all of the samples were calculated by weight loss measurement. The samples were immersed in 3.5 wt % NaCl solution for 7 days. Corrosion products were removed by immersing in a chromic acid cleaning solution (200 g/L CrO3 + 2 g/L AgNO3) at 25 °C for 5 min.

Table 2

Electrochemical Parameters Obtained from the Polarization Curves and the Corrosion Rate (CW) (Evaluated from Weight Loss Measurement by Using eq )

samples	Ecorr (V vs SCE)	icorr (μA cm–2)	–βc (mV/dec)	CW (mm y–1)
AZ31	–1.552 (±0.008)	249.4 (±5.2)	143.4 (±0.3)	2.35 (±0.06)
as-rolled	–1.529 (±0.005)	148.2 (±4.7)	143.9 (±0.4)	1.51 (±0.08)
AT 400	–1.518 (±0.006)	108.6 (±4.2)	103.8 (±0.9)	1.23 (±0.06)
AT 200	–1.491 (±0.009)	32.10 (±3.8)	115.6 (±0.7)	1.05 (±0.07)

Electrochemical Measurements

Open-Circuit Potential (OCP)

Figure shows the open-circuit potential (OCP) curves of all of the samples immersed in 3.5 wt % NaCl solution for 1800 s. It is reported that an increasing OCP value represents the initiation and propagation of corrosion. Meanwhile, a relatively stable OCP value indicates the advance of corrosion and the deposit of corrosion products in a steady state.[24] It can be seen that the OCP values of all of the samples maintained a relatively stable state after immersion for 1800 s. The stable OCP values of the sample can be ranked as AZ31 < as-rolled < AT 400 < AT 200. Compared with the as-extruded AZ31 plate after the solution-treated sample, the AT 200 sample exhibited a higher stable value of ∼12 mV (vs SCE). A previous study has demonstrated that the formation of a more protective surface layer on the Mg alloy led to a higher OCP value.[25] However, slight fluctuations of the OCP values were probably caused by local pitting corrosion.[26]

Figure 7

Open-circuit potential (OCP) versus time curves of the as-received AZ31, as-rolled, AT 400, and AT 200 samples immersed in 3.5 wt % NaCl solution for up to 1800 s.

Potentiodynamic Polarization Curves

The potentiodynamic polarization curves of as-received AZ31, as-rolled, AT 400, and AT 200 alloy sheets were obtained in 3.5 wt % NaCl solution as shown in Figure . The anodic and cathodic branch curves represent the dissolution of the magnesium matrix and the hydrogen evolution, respectively.[27] The electrochemical parameters were determined from the Tafel extrapolation method, and the results are summarized in Table . Generally, smaller icorr and positive Ecorr represent a lower dynamic corrosion rate and thermodynamical corrosion tendency, respectively.[28,29]

Figure 8

Potentiodynamic polarization curves of as-received AZ31, as-rolled, AT 400, and AT 200 samples.

Carefully inspecting the data, the Ecorr and icorr of as-received AZ31 alloy sheets were 1.552 V (vs SCE) and 2.49 × 10–4 A cm–2, respectively. After rolling seven passes, the Ecorr of the sample increased to 1.529 V (vs SCE) and the icorr decreased to 1.48 × 10–4 A cm–2. The corrosion resistance of as-rolled alloy sheets was enhanced because of a stronger basal texture, which is well consistent with previous reports.[9,30] Compared with the AT 400 sample, the lower corrosion resistance of the as-rolled alloy sheets was mainly caused by the higher dislocation density (Figure b).[31] With the increase of dislocation density, the local potential tends more to disequilibrium, leading to the increase of the dissolution rate of the anode metal.[32] While subsequently annealing at 200 °C, the Ecorr tends to become more positive (1.491 V vs SCE) and the icorr further reduced to 3.21 × 10–5 A cm–2 (AT 200). It is clearly found that the grain size and dislocation density in the alloy sheets decrease because the SRX occurred after annealing treatment (Figure d). In addition, the second phase in the Mg matrix, which acts as a microcathode, could decrease the corrosion resistance of the Mg substrate.[3] In this study, few second-phase particles were found inside as-received AZ31 and AT alloy sheets, so that the effect of Mg17Al12 on the corrosion rate was not considered. The electrochemical polarization results showed a good agreement with hydrogen evolution and weight loss measurement results.

Electrochemical Impedance Spectroscopy (EIS)

Generally, EIS is applied to evaluate the corrosion behavior of the samples. Figure shows the EIS curves of as-received AZ31, as-rolled, AT 400, and AT 200 samples measured in 3.5 wt % NaCl solution after 1800 s stabilization. Obviously, the Nyquist plots of four samples all consisted of a capacitive loop in the high- and medium-frequency regions and a low-frequency inductance loop, as labeled in Figure a. It could be clearly seen that the diameter of the capacitive loop increased gradually after rolling and subsequent annealing, which indicated that the AT 200 sample exhibited the best corrosion resistance. In general, the presence of a high-frequency capacitive loop could be attributed to the feature of a double electric layer in the electrochemical reaction process, which corresponds to the degradation of the Mg matrix. The inductance loop at the low-frequency region was related to corrosive pitting and the influence of corrosion products produced during the corrosion process.[3,33] Meanwhile, the Bode modulus plots of impedance vs frequency are shown in Figure b; it could be found that the impedance value of AT 200 alloy sheets in the low-frequency region was higher than the other alloy sheets. As shown in Figure c, the Bode plots of phase angle vs frequency consist of two time constants; the one in the low-frequency range corresponds to the resistance of the corrosion product film, and the other in the high-frequency range results from the double-layer capacitance and the corresponding charge transfer resistance in the presence of corrosion products.[20] Besides, the two wave crests imply two capacitance loops, and the single wave trough occurring in the low-frequency range represents the inductance loop.

Figure 9

Electrochemical impedance spectra (EIS) of as-received AZ31, as-rolled, AT 400, and AT 200 samples measured in 3.5 wt % NaCl solution after 1800 s stabilization, (a) Nyquist plots, (b) Bode modulus plots, (c) Bode plots of phase angle, and (d) equivalent circuit models used for fitting the electrochemical impedance spectra for the four alloy sheets.

To further clarify the corrosion mechanism of the above alloy sheets, the EIS spectrum was simulated, and the appropriate electrochemical equivalent circuit is depicted in Figure d. Here, Rs represents the solution resistance; Rct is charge transfer resistance; and RL and L represent resistance and inductance, which are used to describe the low-frequency loop and the initiation of localized corrosion, respectively.[34] The constant phase element (CPE), Q, is used to represent the double-layer capacitance (Cdl); the calculation of Cdl can be estimated by using Brug’s formula[35−37]where Q is the CPE value (Ω–1 s cm–2) and n is the CPE dimensionless exponent. If n is approximately equal to 1, the Q is identical to an ideal capacitor.[38] The corresponding fitting results are listed in Table ; it can be found that the Rct value of the AT 200 sample is approximately 3 times that of the as-received AZ31 plate. The Cdl values were 10 and 3.52 μF cm–2 for the AZ31 and AT 200 samples, respectively. Generally, the higher values of Rct indicate the lower degradation of the Mg substrate, and the lower Cdl values imply a more compact surface.[16,24] Therefore, hot-rolling and subsequent annealing are beneficial to enhance the corrosion resistance of as-received AZ31 alloy sheets, which is well consistent with the results of the polarization curve test.

Table 3

Fitting Results of the EIS for the Four Samples

sample	Rs (Ω cm2)	Rct (Ω cm2)	Qdl (Ω–1 sn cm–2)	n	Cdl (μF cm–2)	RL (Ω cm2)	L (H cm–2)
AZ31	10.02	220.3	1.363 × 10–5	0.968	10.0	151.6	3.08
as-rolled	10.50	335.5	1.311 × 10–5	0.961	9.04	196.4	1.25
AT 400	10.15	460.9	1.113 × 10–5	0.935	4.26	254.2	2.25
AT 200	12.39	592.5	9.183 × 10–6	0.912	3.52	338.1	2.85

To further clarify the initiation of corrosion and the role of grain boundaries in the corrosion process, as-received AZ31, AT 400, and AT 200 alloy samples were processed by electropolishing and then immersed in 3.5 wt % NaCl solution for 30 min. The EBSD test was conducted after removing corrosion products; the IPF maps and the corresponding band contrast images of the three samples are displayed in Figure . As shown in Figure a, a large corrosion area (black area) was observed on the surface of the AZ31 alloy with a grain size of ∼22 μm, which represents the corrosion area and H2 evolution. After annealing at 400 and 200 °C, the grain size of the alloy decreased to ∼11 and ∼4 μm, and the corrosion area on the samples reduced, as can be seen in Figure b,c. Meanwhile, from Figure a–c, the initiation of corrosion mainly occurs inside grains and then gradually expands with prolonging immersion time. This indicated that the grain boundaries as a physical barrier played an important role in the corrosion process, leading to the propagation of local pitting corrosion being suppressed in the grain refinement alloy sheets. Hence, smaller grain sizes increased the grain boundary area. As a consequence, the corrosion resistance of the alloys with smaller grain sizes was obviously improved.

Figure 10

IPF maps of (a) as-received AZ31, (b) AT 400, and (c) AT 200 samples immersed in 3.5 wt % NaCl solution for 30 min. Corrosion products were removed by immersing in chromic acid solution (CrO3 200 g/L + 2 g/L AgNO3) and the corresponding band contrast image (d–f), respectively.

In addition, after immersing in 3.5 wt % NaCl solution for 7 days and then removing the corrosion products, the corroded surface morphologies of the alloys are shown in Figure . As for the as-received AZ31 alloy, it corroded seriously, with very large and deep corrosion pits observed on the surface, as can be seen in Figure a. After rolling seven passes, some localized corrosion still existed, as shown in Figure b. However, no obvious corrosion pits were seen on the surface of the AT 200 sample, as shown in Figure d. Therefore, these phenomena further demonstrated that the hot-rolling and subsequent annealing are beneficial to improve the corrosion resistance of the as-received AZ31 alloys.

Figure 11

Surface morphology of (a) as-received AZ31, (b) as-rolled, (c) AT 400, and (d) AT 200 samples after immersion in 3.5 wt % NaCl solution for 7 days. Corrosion products were removed by immersing in chromic acid cleaning solution (CrO3 200 g/L + 2 g/L AgNO3) at 25 °C for 5 min.

To better understand the influence of grain sizes on the corrosion behavior of alloy sheets, the schematic illustration is provided in Figure . In the initiation stage of corrosion, the corrosion of alloy sheets with both large and small grain sizes occurred inside of the grains (Figure ). As shown in Figure d, no obvious corrosion pits appeared, and the corrosion morphology mainly exhibited shallow corrosion. As for the alloy sheets with larger grains, more large grains were gradually corroded and resulted in the formation of corrosion pits. With the extension of immersion time, the corrosion area was enlarged and the depth of corrosion pits was deeper than that of the uniformly refined grain alloy sheets.

Figure 12

Schematic illustration showing the corrosion process of (a) as-received AZ31 and (b) AT 200 samples.

Conclusions

The microstructures of the as-received AZ31 alloy sheets were homogenized by hot-rolling and subsequent annealing, which caused obvious grain refinement, few second-phase particles, higher fraction of grain toward basal orientation, and a lower dislocation density. Moreover, the ultimate tensile strength (UTS), yield strength (YS), and fracture elongation (FE) increased from 232 to 273 MPa, 124 to 205 MPa, and 12.5 to 26.4%, respectively, mainly attributed to more uniform grain size and weakened basal texture after static recrystallization. The corrosion resistance of the as-received AZ31 alloy was improved by the hot-rolling and subsequent annealing, which can be mainly ascribed to the more uniform refined microstructure and the lower dislocation density. The EBSD results of the corrosion morphology revealed that the initiation of corrosion mainly occurred inside of grains and the propagation of local pitting corrosion was suppressed by grain boundaries.