Fatemeh Iranshahi,Mohammad Bagher Nasiri,Fernando Gustavo Warchomicka,Christof Sommitsch

Institute of Materials Science,Joining and Forming,Graz University of Technology,Kopernikusgasse 24/I,Graz 8010,Austria

Received 16 March 2020;received in revised form 27 June 2020;accepted 9 August 2020 Available online 2 October 2020

Abstract The increasing use of light alloys owing to their high performance makes magnesium alloys very attractive for the use in automotive and biomedical applications.However,it is well known that magnesium and its alloys have poor corrosion resistance in different atmospheric and aqueous environments.As a means of improving corrosion resistance through the microstructure modification,electron beam processing(EBP)was applied on the as-cast AZ91 magnesium alloy.To evaluate the microstructure influence on the corrosion-resistant,the EB processed samples underwent a solution heat treatment and an artificial aging heat treatment.Four different obtained microstructures were investigated by standard microscopy and electrochemical corrosion tests to evaluate the microstructure and its effects on the corrosion resistance of AZ91 alloy.The EBPed specimens show a significant microstructure refinement and homogenous distribution ofβ-phase at the grain boundaries surrounded by supersaturatedα-Mg which acts as a barrier against corrosion.The electrochemical corrosion test of the samples immersed in 3.5 wt% NaCl after 4 weeks indicates that the EBP improves the corrosion resistance of the alloy due to the nobler corrosion potential of supersaturatedα-Mg and more stable protective hydroxide films compared to the heat-treated and as-cast conditions.

Keywords:Magnesium alloys;Electron beam processing;Corrosion;AZ91;Microstructure;Grain refinement.

1.Introduction

Among the lightweight metallic alloys,magnesium alloys owing to their distinctive properties such as high strength to density ratio and excellent castability are suitable alloys for automotive,aircraft,and aerospace industries[1].In magnesium alloys,aluminum is the most commonly used alloying element due to several beneficial effects such as increased strength,hardness,and castability[2].The second most effective and commonly used alloying element is zinc,which increases room-temperature strength in combination with aluminum[3].Manganese is used as an alloying element to improve the corrosion properties of Mg alloy by removing iron from the melt[4,5].Among aluminum-containing magnesium alloys,AZ91 is a typical alloy that has been widely used in industries.Since Mg is the most active engineering material,its application in general and thus the applications of AZ series of magnesium alloys are limited due to their rapid corrosion rate[6–8].To increase the corrosion resistance,understanding the corrosion mechanisms of the different microstructure is crucial[7].The microstructure of AZ91 is mainly determined by the solidification and cooling rate experienced during the processing[9].The solidification of AZ91 alloy starts first with nucleation of primaryα-Mg in the range of 650–600°C[10].Growing the primaryα-Mg is associated with the accumulation of the alloying elements in the interdendritic liquid phase,which can reach the eutectic composition under non-equilibrium solidification conditions.Further under cooling stimulates the enriched molten phase to solidify in the forms of(α+β)eutectic[11,12].The eutectic transformation in AZ91 alloy is of great importance since it controls the size,shape,and distribution ofβ-phase which in turn influences the ductility,creep,and corrosion resistance[13].AZ91 alloy shows fully or partially divorced eutectic:a fully divorced morphology comprises supersaturatedα-Mg nucleated from primaryα-Mg which surrounds the eutecticβparticles,while the partially divorced morphology is characterized by formingα-Mg islands within theβ-phase from the depleted melt in the last stages of solidification[12–14].The partially or fully divorced eutectic transformation mainly depends on solidification or cooling rate and by increasing the cooling rate,the microstructure tends to be more divorced[13,15].The eutectic is metastable in the AZ91 alloy and thus under a sufficiently low cooling rate of casting the supersaturated eutecticα-Mg decomposes to the lamellar structure ofα-Mg andβ-Mg17Al12by discontinuous precipitation[12,13].

The corrosion rate of AZ91 is influenced by the mutual effect ofαandβ-phase,their morphology,and content of aluminum inα-Mg[9,16,17].Microgalvanic coupling betweenα-Mg andβ-Mg17Al12accelerates the corrosion rate ofα-Mg which is mainly due to more negative free corrosion potential ofα-Mg(?1.6V)than that ofβ-Mg17Al12(?1.3V)[18–20].The corrosion potential difference between the two phases varies by changing the Al content inα-Mg.Generally,an increasing amount of aluminum inα-Mg results in an overall decreasing corrosion potential difference betweenα-andβ-phase.Grimm et al.reported a relatively stable corrosion potential of singleα-phase up to 8wt.-% Al and an exponential rise of potential with increasing Al-content up to the solubility limit at 12.7wt.-%.Al[7].It is worth noting that the protective effect of corrosion products is also sensitive to Al content.According to several research works[19,21–23]the significant drop of the corrosion rate happens at 4wt.-%Al-content which can be attributed to a transition in oxide stability.In the dual-phase microstructure,the intermetallicβ-Mg17Al12phase has contradictory roles affecting the corrosion behavior of AZ91.In the case thatβ-phase forms a more or less continuous network,it acts as a barrier against the corrosion,while coarse and isolatedβparticles accelerate the corrosion by acting as a cathode[7,17,22,24,25].Furthermore,Grimm et al.showed that a fully interconnected supersaturatedα-phase surrounding the primaryα-andβ-phase could act as an effective corrosion barrier despite low continuity ofβ-phase[7].Therefore,microstructure modification with controlling the grain size,volume fraction,size,and distribution of the secondary phases can be one of the strategies to control the corrosion of the magnesium alloys.

One of the possible microstructure modification approaches is microstructure refinement,which can lead to the improvement of the mechanical and corrosion properties.Generally,a refined microstructure obtained from a rapid solidification process can improve the corrosion properties[26,27].Supersaturation achieved due to the rapid solidification leads to an increase of nobler elements in solid solution in the matrix.The nobler elements(i.e.Al)in Mg solid solution increase the corrosion potential and enhance the stability of the oxide layer in magnesium alloy.Moreover,the more homogeneous distribution of the microstructural elements due to the rapid solidification lowers the negative effect of impurities,such as sensitivity to pitting corrosion[19].However,the precipitation of the noble elements and matrix can form microgalvanic cell which decreases the corrosion resistance.Also,grain refinement through the rapid solidification with a sufficient fraction of the intermetallic phases could provide a passive condition[22].

Different techniques are used for rapid solidification such as laser surface melting[28],centrifugal atomization[19],and electron beam melting[29].The improvements of wear and corrosion resistance of magnesium alloys after laser melting have been reported in several investigations[30,31].Abbas et al.studied the effect of laser melting on corrosion resistance of magnesium alloys AZ31,AZ61,and WE43,and their results indicate a significant improvement in corrosion resistance which is a result of grain refinement and uniform redistribution of the intermetallic phases[30].Dubéet al.reported an enhanced corrosion rate of AZ91D and AM60 alloys after laser surface melting using a pulsed Nd-YAG Laser[32].In addition to the laser surface melting technique,electron beam processing(EBP)is one of the most precise techniques with very high efficiency for the surface modification of materials.In this method,an electron gun composed of a tungsten cathode and an anode placed in a high-vacuum generates a dense stream of high-velocity electrons to bombard,heat,and melt the workpiece[33].EBP exhibits essential advantages over laser and ion beams with its great efficiency,simplicity,and reliability[34].Recently,Liu et al.found that Mg-4Sm alloy treated by high current pulsed electron beam(HCPEB)shows the best corrosion resistance,which can be mainly attributed to the homogenous microstructure and composition after repeated melting and rapid solidification[35].The present work investigates the influence of the electron beam processing on the microstructure of AZ91 alloy and its corrosion behavior.To understand the corrosion mechanism as a function of microstructure,a comparative study has been planned to investigate the corrosion characteristics of AZ91 alloy in as-cast,EB processed,and heat-treated conditions.The key aspect in the present work is the characterization of the microstructure of AZ91 alloy developed by EBP and the evaluation of the role played by the microstructure constituents,their size and morphology,and their mutual effects on corrosion rate.

2.Experimental

2.1.Material

The material used in this work was a 10mm thick plate of as-cast AZ91 magnesium alloy produced by nonferrum metal powders GmbH(St.Georgen,Austria),with a chemical composition of Mg(balance),8.5 wt% Al,0.76 wt% Zn,0.21 wt% Mn,0.0008 wt% Ni,0.0049 wt% Cu,0.04 wt% Si and 0.002 wt% Fe.

2.2.Electron beam processing

Alloy microstructure modification was performed on the AZ91 plates using an electron beam machine model Probeam EBG 45–150 K14(Probeam GmbH & Co.KGaA,Germany).To find the optimized condition of EBP,a series of preliminary tests were carried out in a vacuum chamber(5×10?7Torr),with beam current in the range of 6–10mA,and beam speed in the range of 10–20mm/s using a circle beam figure with a diameter of 2.5mm–5mm.The effects of the EBP parameters on the surface quality and the depth of the molten zone were studied to find the optimal conditions.All the rest experiments were implemented in the optimized condition.

2.3.Heat treatment

To investigate the effect of the microstructure constituents including ofα-Mg,β-phase and supersaturatedα-Mg together with the effect of phase morphologies on corrosion behavior of the processed alloys,two heat treatments were planned to be carried out on electron beam processed(EBPed)specimens comprising solution treatment(HT 4 samples)at 413°C for 16 h and artificial aging treatment(HT5 samples)at 216°C for 4 h,according to ASTM B661-12 standard[36].

2.4.Microstructural characterization

The specimens for microstructure observation were ground with SiC papers up to P4000,polished by diamond paste up to 1μm and final step with colloidal silica solution(OPS)of 0.04μm,followed by rinsing with ethanol.Then the samples for optical microscopy were etched by an acetic-picral solution(2.1 gr picric acid+5ml acetic acid+5ml distillate water+70ml ethanol)for only 1 second.To investigate the specimens after corrosion tests,corroded samples were cleaned with chromic acid to remove the corrosion product from the surface of the samples and then rinsed using a large amount of water.The microstructure characterization was performed with a Zeiss observer Zm1 optical microscope(OM)and a field emission scanning electron microscope(FESEM)model TESCAN MIRA3 with energy-dispersive X-ray spectroscopy(EDS)and electron backscattered diffraction(EBSD).To evaluate the grain size of EBPed specimens,the EBSD maps were analyzed by OIM software.As well as,the average grain size of the as-cast alloy was determined by the linear intercept method according to the ASTM E112-12 using OM images[37].

2.5.Electrochemical measurements

Corrosion behavior of the specimens was investigated by open circuit potential(OCP)variation with time,potentiodynamic polarization(PDP),and electrochemical impedance spectroscopy(EIS)using an Autolab(PGSTAT128N)potentiostat/galvanostat.A three-electrode cell including a counter electrode made of platinum foil,a reference electrode made of saturated Ag/AgCl,and with the specimens as the working electrodes were used for electrochemical measurements.Rectangular specimens with only one side exposed were embedded into epoxy resin,were ground up to P4000 SiC papers,washed with distilled water,and then dried using a hairdryer.The measurements were carried out in a 3.5% NaCl solution at room temperature.Before each electrochemical measurement,the OCP variation with time was measured at least 45min to attain a steady-state.The potentiodynamic polarization(PDP)tests were performed in the potential range of±30mV from OCP with a scan rate of 1mV/s.EIS analysis was performed in a frequency range of 0.01Hz to 100kHz using the amplitude of the sinusoidal potential signal of 10mV for immersion time intervals from 1h to 4 weeks.At least three measurements were performed for each condition.Afterward,the results of the measurements were analyzed using NOVA1.11 software.

3.Results and discussion

3.1.Parameter optimization of the process

The surface quality and the depth of the processed area along with defect-free molten zone are the criteria to evaluate the quality of the EBP.The EBP is being controlled by the accelerating voltage(U),current of the beam(I),speed of the beam(υ),focusing current,and the spot size(?)[38].Several EBP beads on plates with different processing parameters were examined to reach the most appropriate processed area on the alloy.The processed samples were compared based on the existence of defects(e.g.,pores,cracks)and quality of the surface.The parameters of the EBP producing samples without any defects and smoother surface appearance were selected as optimized parameters which are summarized in Table 1.

The energy input(Q)is the main factor that determines the width and depth of the molten zone which is calculated based on:

whereQis energy input;Uis process voltage;Iis beam current;vis beam travel speed.The cross-section and top view of the single bead on the plate using those optimized parameters are shown in Fig.1.

Increasing the penetration of the electron beam is required to perform the electrochemical corrosion test on the processed specimen,hence the energy input increased from 45J/mm to 150J/mm to reach a melted area with the depth of 5.66mm free of any defects(Fig.1(c)).

3.2.Microstructure characterization

As shown in Fig.2,the as-cast microstructure contains primaryαas a matrix surrounded by decomposedαeutectic.Theβeutectic islands are observed among the decomposedαeutectic region.The decomposedαeutectic regions comprise lamellarβprecipitated from supersaturatedαeutectic.Also,some discrete Al–Mn particles with sharp edges are observed mostly in the middle of the grains.Since the melting temperature of Al–Mn compounds is above 700°C,insoluble Al–Mn intermetallic particles are acting as potent nucleation sites for bothα-andβ-phases during the casting process[10,39].

Table 1Electron beam parameters used for microstructure modification of AZ91.

Fig.1.Cross-section of EBP single bead on AZ91 plates using different parameters(a)test no.1,(b)test no.2(c)test no.3,and(d)top view of test no.3.

Fig.2.Micrographs of as-cast AZ91 by(a)OM and(b)FESEM(BSE Mode).

Fig.3.(a)FESEM image,EDS(b)point,and(c)Al-map analysis of EBPed alloy.

Table 2EDS point analysis of EBPed alloy.

Fig.3 displays the microstructure of the EBPed AZ91 alloys analyzed by the backscattered electron(BSE)mode of FESEM,EDS point,and mapping analysis of the EBPed specimen for aluminum distribution.The microstructure of the EBPed specimens represents a homogenous distribution of theβ-phase in theα-Mg matrix and randomly distributed Al–Mn particles.Due to the high cooling rate during EBP,decomposition of supersaturated eutecticα-Mg to the lamellarβ-phase has not been observed compared to as-cast condition(Fig.2).As shown in Fig.3,EBPed AZ91 has a refined microstructure compared to the as-cast condition shown in Fig.2.The data of EDS analysis of three spots in Fig.3b are given in Table 2.According to the EDS data,the lower Al-containing areas are darker than higher Al-containing places.EBPed microstructure shown in Fig.3b consists of dark primaryα-Mg(～4 wt% Al.)area surrounded by a brighter network of supersaturated eutecticα-Mg(～11 wt% Al.)in direct contact with fully or partially divorcedβ-Mg17Al12.

Fig.4 shows the inverse pole figure(IPF)map obtained from the EBSD analysis of EBP AZ91.Compared to as-cast,the microstructure of EBPed is characterized by finer grain,finer divorcedβ-phase,and a connected network of supersaturatedα-Mg.The average grain size of the alloy decreases from 280μm in as-cast condition to 15μm in the EBP condition.

The microstructure of EBPed AZ91 specimens after heattreatment(HT4 and HT5)is shown in Fig.5.Solution heat treatment(HT4)of the EBPed alloy at 413°C led to the almost complete dissolution of the secondary phases and aluminum homogenization in theαmatrix,while the Al-Mn particles remain(Fig.5(a)and(b)).In the artificial aging heat treatment(HT5)of EBPed AZ91 specimens,very fine and plate shapeβ-phase particles are discontinuously precipitated from the supersaturatedα-Mg around the eutecticβ-Mg17Al12phases(Fig.5(c)and(d)).EDS analysis shows the aluminum content of theα-Mg matrix is around 8 wt% in HT4 samples.Due to the decomposition of supersaturatedαeutectic in HT5 condition,aluminum content in this area decreases to around 4 wt%.

Fig.4.Inverse pole figure(IPF)map of EBPed condition.

3.3.Corrosion behavior

3.3.1.Open circuit potential(OCP)

The alloy in different conditions was exposed to 3.5 wt%NaCl for 2700 s to determine the open circuit potential(OCP)as a function of time and measurement was repeated three times for each case.Fig.6 shows the mean values of OCP with time after the immediate immersion of the specimens into the solution.As seen,EBPed alloy possesses a higher OCP than those under the other conditions.In general,the increasing OCP after immersion is due to the growth of the corrosion layer,whereas a dynamic balance between the advance of corrosion and the deposit of corrosion products leads to a relatively stable OCP value[40].The OCP of as-cast and HT5 conditions shows a similar trend,which is in agreement with the similarity of their microstructures,as shown in Figs.3 and 5.In as-cast condition,the OCP increases within the initial 300 s of immersing toward the positive direction from?1.62V to?1.60V which is related to the formation of the magnesium hydroxide film on the surface[1].Subsequently,the potential slightly increases,and for the last 1200 s is stabilized at around?1.58V.The fine microstructure of HT5 condition,with dispersed fineβ-phase,facilitates the formation of a more stable hydroxide film compared to as-cast condition.The OCP in HT5 condition thus stabilizes much earlier compared to the as-cast condition and its potential is slightly higher than the as-cast condition.The OCP in the HT4 condition increases fast in the first 120 s,indicating the formation of the magnesium hydroxide film.It subsequently decreases to the initial value indicating that the hydroxide film is less stable[41]in the absence ofβ-phase despite higher Al content in theα-Mg matrix.The less stable and protective hydroxide film in HT4 condition together with a lower potential compared to the as-cast condition shows the effectiveness ofβ-phase in microstructure acting as a barrier to avoid further corrosion.Generally,the presence of theβ-phase in microstructure increases the free potential in a positive direction[42],[43].EBPed alloy shows a significant continuous increase of potential during the first 1500 s from?1.64V to?1.52V.It can be attributed to the corrosion of primaryα-Mg.Due to the less Al content of primaryα-Mg,its corrosion potential is more negative and the composed hydroxide film on top of that is less protective[7],therefore primaryα-Mg is corroded faster than the supersaturatedα-Mg.By corrosion of the primaryα-Mg,the remained interconnected supersaturatedα-Mg acts as an interface between the interior primaryα-phases and corrosion media.Therefore,by decreasing the volume fraction of primaryα-phase in contact with corrosion media the corrosion potential increases continuously in the first 1500 s.On increasing the time,the potential dropped to the negative values(?1.54V)due to the dissolution of the hydroxide film attacked by Cl?ions and gradually stabilized at about?1.56V for the last 600 s of immersion,indicating that passivation with the corrosion products has occurred[44].

3.3.2.Potentiodynamic polarization

The potentiodynamic polarization curves depicted in Fig.7 reveal a rapid increase of anodic curve on increasing the potential.The EBPed sample shows a significant shift of the polarization curve to the smaller current density,which is nearly 10 times less than the as-cast condition.Additionally,the corrosion potential of EBPed alloy is slightly more positive than the observed for the as-cast condition.The similar current density in as-cast and EBP-HT5 conditions is in agreement with their similar microstructure.

Fig.7.Potentiodynamic polarization curves of as-cast and EB processed AZ91 immersed in 3.5% wt NaCl solution.

Electrochemical parameters of potentiodynamic polarization were calculated by Tafel extrapolation of the cathodic branch of the polarization curve,as listed in Table 3 for all the conditions.According to the calculated values,the corrosion potential(Ecorr)of the as-cast alloy is?1.460V whilst,noblerEcorrof?1.435V determined for the alloy after EBP.Moreover,the EBPed alloy shows lower corrosion current density(icorr;4.38μA/cm2)than that of as-cast condition(39.03μA/cm2).NoblerEcorrand lower corrosion current density of EBPed alloy can be attributed to the less galvanic potential between supersaturatedα-phase andβ-phase compared toα-phase andβ-phase in the as-cast condition.The EBPed-HT4 shows less corrosion current density compared to the as-cast condition.This can be attributed to the absence of a microgalvanic reaction since theβ-phase does not exist in the microstructure of EBPed-HT4 samples.Higher values of anodic slope generally indicate the occurrence of a cathodic controlled reaction for all the samples[1].The corrosion rate of Mg alloy(mm/year)can be estimated with the corrosion current density,icorr(mA/cm2)estimated by Tafel extrapolation using the Eq.(2)[45].Although Song and Atrens reported[16]that the corrosion rate calculated using Tafel extrapolation data can be reliable only for the short-term or transient estimation[46],however,it is widely used to evaluate the corrosion of Mg alloys since it is quick and easy to evaluate and compare the effect of different conditions[45].

The metal weight loss measurement is a simple way to evaluate the overall corrosion rate using Eq.(3)as proposed in[45]and[47].The weight loss(g/cm2/d)of the immersed samples in 3.5 wt% NaCl was measured after a week of immersion for all conditions.The estimated corrosion rates(mm/year)using Eq.(3)are given in Table 3.In contrast to the corrosion rate calculated using Tafel extrapolation data,the corrosion rate estimated by weight loss indicates long term corrosion rate.Identical toPi,AZ91 alloy in EBPed condition has a lowerPwcompared to the other conditions while the corrosion rate for EBPed-HT4 is the maximum among the all conditions.

It is worth noting that theEcorris not the only effective parameter on the corrosion rate and as shown in Table 3,Ecorrand corrosion rate do not show a direct correlation.

Table 3Electrochemical parameters of potentiodynamic polarization calculated by Tafel extrapolation.

3.3.3.Electrochemical impedance spectroscopy

Since the anodic dissolution rate is fast and the anodic Tafel slope shows non-polarizable characteristics,the determination of the long-term corrosion rate through the Tafel extrapolation for Mg alloy is complicated and inaccurate,since the anodic dissolution rate is fast,and the anodic Tafel slope shows non-polarizable characteristics[48].Hence,electrochemical impedance spectroscopy(EIS)is an efficient continuous monitoring approach to understand the corrosion mechanism and investigate its kinetics[25].Fig.8 shows the Nyquist plot of as-cast,EBPed,and heat-treated EBPed AZ91 alloys up to 4 weeks(672h)of immersion.The Nyquist plots for all the samples consist of a capacitive loop at high-and medium-frequencies and an inductive loop at low-frequencies.The high-frequency part in the Nyquist plot represents the charge transfer resistance[49]showing the oxide film properties,whereas the low-frequency part indicates the occurrence of Faradaic processes on the metal ascribing the relaxation of the adsorbed species[49,50].The capacitive behavior in the high-to the medium-frequency region refers to the resistance of the oxide film.In the early stages of immersion(Fig.8a),the EBPed-HT5 and as-cast samples show a better capacitive behavior,indicating the formation of a more resistive oxide film compared to the other conditions.This can be attributed to the higher volume fraction ofβ-phase in their microstructure with high aluminum content.Due to the similar microstructure,the EBPed-HT5 and as-cast conditions show similar electrochemical impedance responses which are in agreement with potentiodynamic polarization and OCP test.However,after 672 h(Fig.8.d),the oxide film formed on the EBPed-HT5 sample is more corrosion resistant than oxide film formed on the as-cast condition,attributed to the more continuity ofβ-phase due to the finer microstructure of EBPed-HT5.

The oxide film formed on the EBPed alloy shows higher corrosion resistivity compared to other conditions after 24 h.Better capacitive behavior of EBPed sample after 24 h indicates that the primaryα-phase has been corroded and supersaturatedα-phase with higher aluminum content is the dominant phase in contact with the electrolyte.The EBPed-HT4 sample shows the inferior capacitive behavior due to the absence ofβ-phase in its microstructure.

Table 4Electrochemical parameters of electrochemical impedance spectroscopy.

In addition to the qualitative interpretation of the Nyquist plot,a Randles equivalent circuit consisting of a resistor connected in series to a parallelly connected resistor and a capacitor has been used to fit the EIS experimental data.The physical counterpart of the Randles equivalent circuit is the resistance of the electrolyte and surface film(Rs),charge transfer resistance of the metal(Rct),and the electrical double layer capacitance at the metal/solution interface(Cdl).Electrical double-layer capacitance shows the water/electrolytes uptake through pores/defects of the hydroxide film on the surface[42,43],thus,an increase ofCdlwith time implies the increase of porosities/defects in the hydroxide film.The corresponding electrochemical parameters are presented in Table 4.As shown in Table 4,EBPed-HT5 and as-cast samples show higher charge transfer resistance(Rct)in the early stage of immersion(1 h)whereas,in longer immersion time,EBPed shows higherRct.Increment of the charge resistance by increasing immersion time implies the sequential growth of the protective hydroxide layer at the alloy surface.The observed fluctuation inRctvalues can be partly attributed to the undermining of theβ-phase particles[16],however,an increasing trend inRctvalues is observed for all conditions.The estimatedCdlvalues for the as-cast condition are almost twice those of the others indicating the composition of a hydroxide layer with more porosity facilitating diffusionreaction through the hydroxide film.EBPed-HT4 shows lessCdlvalues due to the homogeneous microstructure obtained during the solution treatment.Almost the sameCdlvalues have been measured for EBPed and EBP-HT5 showing almost the same water/electrolytes uptake properties for their hydroxide film.Despite the similar microstructure,the fine microstructure of EBP-HT5 compared to the as-cast condition can be the reason for smallerCdlof EBP-HT5 since a more coherent hydroxide layer with less porosity and defect can be formed on a fine microstructure.

Fig.8.Nyquist spectra in as-cast,EBP,EBP-HT4,and EBP-HT5 AZ91 alloy after(a)1 h,(b)24 h,(c)96 h,and(d)672 h.

3.4.Surface and cross-section of the corroded alloys

The corroded surfaces of the specimens after 1 h,24 h,and 96 h of immersing in 3.5 wt% NaCl were analyzed by FESEM,after cleaning with chromic acid.Fig.9 shows the corroded surface of the as-cast condition.The corrosion in theα-eutectic area is more intensive than in the primaryα-Mg.Sinceα-Mg andβ-Mg17Al12in theα-eutectic area are in contact with together,thus,the microgalvanic corrosion in this region is stronger than that in the primaryα-Mg area.The occurrence of more aggressive corrosion in theα-eutectic area can thus be expected.In further stages,the corrosion attack extends much more in the primaryα-Mg compared to theα-eutectic area as shown in Fig.9c and d.The connected network ofβ-precipitates in theα-eutectic region provides a passive condition,and thus the corrosion extends faster through the primaryα-Mg.Uneven propagation of corrosion through theα-matrix leads to the undermining of precipitated and eutecticβ-phase in further stages of corrosion,as observed in Fig.9e and f.

Fig.9.FESEM images of the corroded surface of the as-cast alloy after(a,b)1 h,(c,d)24 h,and(e,f)96 h.

Fig.10 presents the corrosion morphology of the corroded EBPed specimens after immersion of 1 h,24 h,and 96 h in 3.5 wt%NaCl.Compared to Fig.9,less corrosion is observed in equivalent immersion time.Due to a more negative corrosion potential of the primaryα-phase,the corrosion in EBPed alloy starts from the primaryα-Mg which contains less aluminum compared to the supersaturated area(Fig.10a and b).As seen in Fig.10c and d,the existence of supersaturatedα-Mg with high aluminum content acts as a barrier and decelerates the corrosion rate mostly because of the nobler corrosion potential of theα-eutectic[7].The finer microstructure obtained by EBP shows less undermining of secondary phases,as seen in Fig.10f compared to Fig.9f.It confirms that the supersaturatedα-eutectic in a fine microstructure forms a continually connected network which results in a significant decrease in the corrosion rate and the composition of a more stable passive film.

Fig.10.FESEM images of the corroded surface of EBPed alloy after(a,b)1 h,(c,d)24 h,and(e,f)96 h.

Fig.11 shows the corrosion propagation beyond the surface for both as-cast and EBPed conditions after 24 h immersion.The corrosion of the as-cast is more progressive compared to the EBPed sample with undermining of theβ-phase particles.The localized corrosion in the as-cast specimen indicates weaker passive film in comparison to smooth and homogeneous corrosion in the EBPed specimen,which accords well with theCdlvalues in Table 4.It confirms that the increase of the aluminum in solid solution increases the corrosion potential and enhances the stability of the oxide layer as discussed above[7].

Fig.11.FESEM images of the cross-section of corroded(a)as-cast and(b)EB processed AZ91 after 24 h immersing in 3.5 wt% NaCl solution.

3.5.Corrosion mechanism

To understand the influence of the microstructure on the corrosion behavior of AZ91 alloy,a schematic illustration has been prepared by the conclusion of all experimental results of this research.Fig.12 shows the schematic corrosion behavior of AZ91 alloys with different studied microstructure in the early,intermediate,and long-term immersion.The influence of microstructure on the corrosion resistance can be explained by the cumulative effect of(a)corrosion resistance of the microstructure constituents,(b)corrosion resistance of the hydroxide film(c)the dominant microstructure constituent in contact with corrosion medium.Among the AZ91 microstructure constituents,theβphase is more corrosion resistance with a nobler corrosion potential[19].As discussed in Section 1,the content of Al plays the most important role in the corrosion rate of a singleαphase alloy so that corrosion resistance increases by increasing Al-content.The Al-content also affects the stability of the Hydroxide film,protecting the surface of Mg in aqueous solutions[22,51].

According to the Tafel extrapolation(Table 3),the corrosion rate of the EBPed-HT4 condition is lower than EBPed-HT5 and as-cast condition in the early stage of immersion which is because of higher Al-content ofαphase and insignificant microgalvanic corrosion due to the absence ofβparticles.However,EBPed-HT4 samples show less corrosion resistance according to the EIS test result in the long term of immersion due to the absence ofβparticles that act as a barrier against the extension of the corrosion.Theβparticles decelerate corrosion when they are a dominant phase in contact with the corrosion media.Also,the OCP result implies that the hydroxide film in the EBPed-HT4 condition provides less passivity compared to as-cast and EBPed-HT5.Chelliah et al.[41]showed that AZ91 in HT-4 condition comprised mainly unstable Mg(OH)2film while a passive film containing MgCO3and CaCO3is formed on the surface of as-cast AZ91.Due to the same microstructure of as-cast and EBPed-HT5 conditions in present work,they show almost the same result in all of the corrosion tests.In as-cast and EBPed-HT5 conditions,the corrosion starts from the decomposedαeutectic area which is because of the intensive microgalvanic corrosion between theαphase and the lamellarβphase.In further stages,lamellar and eutecticβphases avoid further corrosion in the decomposedαeutectic area,and thus the corrosion attack extends much more in the primaryα-Mg region(Fig.9).While,in the case of EBPed AZ91 alloy,the corrosion starts from primaryαregion where the Al content is less than the supersaturatedαeutectic region that implies the effect of Al-content in corrosion rate.Generally,an OCP increment is observed for all cases because of the growth of the hydroxide film however this increment is significant just for EBPed samples in agreement with the increment ofRctin the EIS test.Accordingly,increasing the corrosion resistance by time can mostly be attributed to increasing the phase fraction of the supersaturatedαphase andβin contact with the corrosion medium.The SEM images taken from the corroded surfaces imply the effect of theβphase in as-cast condition and supersaturatedαphase in EBPed condition as a barrier against the extension of corrosion.In the as-cast condition,localized corrosion is the dominant corrosion mechanism while EBPed samples are corroded more uniformly(Figs.9 and 10).

Fig.12.Schematic illustration of corrosion mechanism of AZ91 in(a)EBP-HT4(b)as-cast(c)EBP-HT5(d)EBPed conditions.

4.Conclusion

The influence of the electron beam processing on the corrosion resistance of AZ91has been investigated,and the main conclusions can be addressed as follows:

(1)The desired beam penetration in the electron beam processing of the as-cast AZ91 plates with a proper surface appearance and free of volumetric defects has been achieved using the optimized parameters of the EBW machine.

(2)Due to the high solidification and cooling rate during the electron beam processing,a fine microstructure of AZ91 alloy forms with a connected network of supersaturatedα-eutectic and homogeneously distributed eutecticβ-Mg17Al12.

(3)Electrochemical measurements showed that EBPed AZ91 has more positive OCP,lower current density,and better corrosion resistance than those of as-cast mostly because of the positive characteristics of a connected network of supersaturatedα-eutectic including higher potential compared to primaryα-Mg and more stable protective surface films.

(4)The finer microstructure of EBPed and EBPed heattreated AZ91 alloy decreases the undermining process of theβ-phase particles and provides a more coherent hydroxide film.

(5)EIS results in agreement with FESEM micrographs showed that the hydroxide film composed on the surface of a fine microstructure is more protective.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.