Effects of dynamic recrystallization and strain-induced dynamic precipitation on the corrosion behavior of partially recrystallized Mg–9Al–1Zn alloys

Yenny Cubides,Dexin Zho,Lucs Nsh,Digvijy Ydv,Kelvin Xie,Ibrhim Krmn,Homero Cstned,?

aDepartment of Materials Science and Engineering,Texas A&M University,575 Ross Street,College Station,TX 77840,USA

b Department of Engineering Technology and Industrial Distribution,106 Ross Street,College Station,TX 77843,USA

cNational Corrosion and Materials Reliability Laboratory,1041 RELLIS Pkwy,Bryan,TX 77807,USA

Received 30 April 2020;received in revised form 9 August 2020;accepted 6 September 2020 Available online 24 September 2020

Abstract The corrosion susceptibility of recrystallized and un-recrystallized grains in equal channel angular pressed(ECAPed)Mg–9Al–1Zn(AZ91)alloys immersed in chloride containing media was investigated through immersion testing and an electrochemical microcell technique coupled with high resolution techniques such as scanning Kelvin probe force microscopy(SKPFM),transmission electron microscopy(TEM),and electron backscatter diffraction(EBSD).During ECAP,dynamic recrystallization(DRX)and strain-induced dynamic precipitation(SIDP)simultaneously occurred,resulting in a bimodal grain structure of original elongated coarse grains and newly formed equiaxed fine grains with a large volume fraction ofβ-Mg17Al12 precipitates.Corrosion preferentially initiates and propagates in the DRXed grains,owing to the greater microchemistry difference between theβ-Mg17Al12 precipitates formed at the DRXed grain boundaries and the adjacentα-Mg matrix,which induces a strong microgalvanic coupling between these phases.Additionally,the weaker basal texture of the DRXed grains also makes these grains more susceptible to electrochemical reactions than the highly textured un-DRXed grains.The influence of dynamic recrystallization and dynamic precipitation was also studied in ECAPed alloys with different levels of deformation strain through corrosion and electrochemical techniques.Increasing the strain level led to a more uniform corrosion with a shallow penetration depth,lower corrosion rate values,and higher protective ability of the oxide film.Furthermore,higher levels of strain resulted in greater hardness values of the ECAPed alloys.The superior corrosion resistance and strength of the ECAPed alloys with increasing strain level was attributed to the combination of smaller DRXed grain size,higher DRX ratio,and higher volume fraction of uniformly distributed fineβ-Mg17Al12 precipitates.

Keywords:Magnesium alloy;Bimodal grain structure;Dynamic recrystallization;Dynamic precipitation;Severe plastic deformation;Microgalvanic coupling.

1.Introduction

Magnesium is one of the lightest structural metals and has a lower density(1.7g/cm3)than aluminum(2.7g/cm3),titanium(4.5g/cm3),and iron(7.9g/cm3).Therefore,magnesium-based alloys are highly attractive in automotive applications,in which weight reduction improves fuel efficiency and decreases CO2emissions into the atmosphere[1].Similarly,magnesium alloys have also been used in the aerospace industry for reducing weight in several aircraft components[2].Magnesium has also been found to be very attractive in biomedical applications such as orthopedic implants;it has a similar density to that of natural bone(1.8–2.1g/cm3)as well as comparable mechanical properties such as fracture toughness,Young’s modulus,and compressive yield strength[2].Furthermore,magnesium has excellent biocompatibility,and it is nontoxic and biodegradable in body fluids and thus does not require further surgical procedures to remove implants.Finally,magnesium alloys are also promising in electronic applications,owing to their excellent heat transfer and dissipation and shielding effects for electromagnetic and radio frequency interference.

Despite the potential applications of magnesium alloys,some major challenges remain to be overcome.Magnesium alloys have poor ductility and formability,owing to their hexagonal closely packed crystal structure,which limits plastic deformation by slip dislocation[3].They also have lower strength than traditional steel and aluminum alloys.In addition,they exhibit poor corrosion resistance,thus restricting their practical application in exterior environments.Numerous studies have attempted to improve the formability,yield strength,and creep resistance of magnesium alloys,mainly through alloying and processing.Processing routes involving severe plastic deformation and/or heat treatment methods have been extensively developed over the past three decades to produce magnesium alloys that exhibit better mechanical properties than conventional cast structures[4–7].These processing methods lead to small compositional changes as well as microstructural and crystallographic changes such as grain refinement,texture development,changes in dislocation density,redistribution of solute atoms in the microstructure,or precipitation of secondary phases[7–9].These compositional,microstructural,and crystallographic changes can dramatically affect the corrosion behavior of magnesium alloys.However,only limited studies have addressed the roles of these factors on Mg corrosion,and several conflicting findings have been reported.

Grain size is a metallurgical factor that can substantially influence the corrosion performance of Mg and its alloys[10].Grain boundaries are more chemically reactive than the bulk,owing to their higher electron activity and diffusion[11].Therefore,a high density of grain boundaries can promote electron transfer between the metal surface and the electrolyte,thus resulting in more sites for corrosion and consequently in less corrosion resistance with decreasing grain size.However,the higher reactivity of grain boundaries can also provide more sites for oxide film nucleation.Hence,grain refinement can lead to faster formation of an oxide layer and better mechanical adhesion to the underlying metal substrate,thereby increasing corrosion resistance[11].Secondary phases also play a crucial role in the corrosion behavior of magnesium alloys.In Mg–Al alloys such as the AZ91 alloy,β-Mg17Al12precipitates exhibit a dual role in the corrosion behavior in a manner mainly dependent on their volume fraction and distribution.Song et al.[12]have reported that a high volume fraction ofβ-Mg17Al12precipitates,continuously distributed along the grain boundaries,improves the corrosion resistance of Mg–Al alloys by providing a barrier against corrosion propagation.In contrast,a small volume fraction of discontinuousβ-Mg17Al12precipitates accelerates the corrosion rate of Mg–Al alloys by acting as effective galvanic cathodes.Other structural factors such as dislocation density,twinning,and crystallographic texture influence the corrosion behavior of Mg alloys.For example,there is good agreement in the literature that a high density of dislocations increases the rate of anodic dissolution in Mg alloys[3,13,14].Dislocations distort the crystallographic lattice of the metal,and therefore atoms located in these distorted regions are more susceptible to anodic dissolution[13].Similarly,twins are also more energetic sites than the matrix and hence are preferential sites for corrosion[13].Finally,Mg grains with basal orientation have been consistently reported to exhibit higher corrosion resistance than Mg grains with non-basal crystal planes[4,15–17].The closely packed basal plane exhibits a higher corrosion resistance than the more loosely packed non-basal planes because of its higher atomic density(basal(0001)plane=1.13×1019atoms/m2,prismatic(110)plane=6.94×1018atoms/m2,and prismatic(010)plane=5.99×1018atoms/m2)[18].The higher atomic density of the closely packed basal plane is associated with a higher coordination number,a higher binding energy,and a lower surface energy than the loosely packed non-basal planes,thus resulting in an overall higher activation energy for removal of atoms from the metal lattice[11,17].Consequently,the basal plane is more electrochemically stable and usually exhibits the lowest corrosion rate among all other planes in Mg alloys[8,16,18].Further exploration of the combined effects of these microstructural features in improving or deteriorating the corrosion resistance of heavily deformed Mg alloys is vital.In addition,little research has investigated the corrosion susceptibility of recrystallized and un-recrystallized regions in Mg alloys with bimodal grain structures and the combined effect of dynamic recrystallization(DRX)and strain-induced dynamic precipitation(SIDP)on the corrosion behavior of Mg alloys processed though severe plastic deformation methods.

The purpose of this study was to investigate the corrosion susceptibility of recrystallized and un-recrystallized regions in a partially recrystallized AZ91 magnesium alloy processed by equal channel angular pressing(ECAP)in chloridecontaining media.High-resolution methods such as scanning Kelvin probe force microscopy(SKPFM),transmission electron microscopy(TEM),and electron backscatter diffraction(EBSD)were used to elucidate differences in Volta potential,microchemistry,and crystallographic orientation,respectively,between recrystallized and un-recrystallized regions,thus enabling identification of the predominant factors leading to preferential dissolution at specific sites.Furthermore,the microcell electrochemical technique was used to investigate the electrochemical activity of these recrystallized and un-recrystallized regions.The combined effect of dynamic recrystallization and strain-induced dynamic precipitation on the electrochemical behavior of ECAPed alloys at different levels of deformation strain was also analyzed.According to this,microstructural parameters such as grain size of the DRXed grains,DRXed ratio,volume fraction ofβ-Mg17Al12phase,and size ofβ-Mg17Al12precipitates were correlated with the electrochemical response of the ECAPed alloys.Corrosion rate calculations for the ECAPed alloys at different strain levels were obtained through gravimetric techniques.In addition,electrochemical measurements such as open circuit potential(OCP),potentiodynamic polarization and electrochemical impedance spectroscopy(EIS)were used to evaluate the effects of these factors on the corrosion kinetics;a diluted NaCl solution was used to better identify changes in passivation regions and breakdown potentials with increasing strain levels.

Fig.1.Schematic representation of the ECAP process used in this study.1st pass at 310°C;2nd pass at 300°C;3rd pass at 275°C;and 4th pass at 250°C.

2.Materials and methods

2.1.Material preparation

The as-cast AZ91 alloy ingots were supplied by US Magnesium LCC,Salt Lake City,UT,with a chemical composition of 9.2wt% Al,0.67wt% Zn,0.25wt% Mn,and balance Mg.The as-cast ingots were cut and machined into billets of 2.5cm x 2.5cm x 17.8cm.The billets were subsequently solution heat-treated at 413°C for 24h in a protective argon atmosphere,followed by water quenching at room temperature to homogenize the dendritic microstructure of the as-received alloy and to dissolve the majority ofβ-Mg17Al12precipitates.The ECAP processing was performed in an ECAP tool with a 90° die angle and sharp corners.Prior to ECAP processing,the ECAP die and the homogenized samples were preheated to 310°C for 30min to achieve a uniform temperature throughout the entire sample.Samples were extruded for up to four ECAP passes(N=1–4)using routeBc,an extrusion rate of 5mm/min,and a backpressure of 2000psi.The time for each pass including the preheating time,extrusion time,and time for machining the specimen for the next pass was about 2h.The first(N=1),second(N=2),third(N=3),and fourth(N=4)ECAP passes were conducted at 310°C,300°C,275°C,and 250°C,respectively.Following each ECAP pass,the billets were quenched immediately in water at room temperature to prevent grain growth.A schematic illustration of the ECAP processing used in this study is shown in Fig.1.The orthogonal directions after ECAP shown in Fig.1 corresponds to the extrusion direction(ED),the longitudinal direction(LD;normal to the extrusion direction)and the flow direction(FD;orthogonal to both ED and LD).

2.2.Microstructure characterization and hardness measurements

Microstructure analysis of the AZ91 alloys were conducted by optical microscopy(OM,Nikon ECLIPSE MA100)and scanning electron microscopy(SEM;JEOL JMC-6000Plus)equipped with energy dispersive X-ray spectroscopy(EDS,JED-2300).Morphology micrographs of the ECAPed alloys were taken from the LD–ED plane,which is also known as flow plane.Image analysis for calculating average grain size,volume fraction of DRXed grains,and volume fraction and size ofβ-Mg17Al12precipitates were performed using ImageJ software.For the microstructural characterization,samples were wet ground with SiC grinding paper from P240 to P1200,followed by polishing with diamond pastes of 9,3,and 1μm,and finally polished with a colloidal silica suspension of 0.05μm.The polished samples were then etched in an acetic-glycol solution consisting of 1mL HNO3,20mL acetic acid,60mL ethylene glycol,and 19mL water for 10s.The constituent phases of the ECAPed alloys after each ECAP pass were identified using a Bruker powder diffractometer with a Cu-Kαsource(λ=1.54?A=0.154nm)operated at 40kV and 25mA current.The specimens were scanned from 30° to 80° with a step size of 0.03° per second.

Transmission electron microscopy(TEM)was used to compare the morphology and elemental composition difference between the recrystallized and un-recrystallized regions of an ECAPed alloy with fourth ECAP passes.Cross-sectional transmission electron microscopy(TEM)samples were prepared using a FEI HELIOS NANOLAB 460F1 dual-focused ion beam(FIB).The chemical information of the ECAPed alloy was identified using scanning transmission microscopy(STEM)energy dispersive spectroscopy(EDS)mapping on FEI Tecnai G2 F20 Super-Twin FE-TEM operated at 200kV.Electron backscatter diffraction(EBSD,Tescan FERA-3 scanning electron microscope(SEM)with an accelerating voltage of 20kV)was also performed in both the recrystallized and un-recrystallized regions of the ECAPed alloy with fourth ECAP passes to compare their crystallographic orientation.

The Vickers hardness of the polished ECAPed alloys was measured by using a Vickers microhardness tester under a load of 100g for a holding time of 10s.The reported average hardness values were obtained from a minimum of 15 indentation measurements at different locations over the entire sample.

2.3.Localized potential distribution

Volta potential differences between theα-Mg matrix and theβ-Mg17Al12precipitates in the recrystallized and unrecrystallized regions of the ECAPed alloy after the fourth pass was investigated using scanning Kelvin probe force microscopy(SKPFM,Bruker Dimension Icon AFM).Prior to the SKPFM measurements,the ECAPed alloy was mechanically polished as mentioned above,and then it was thoroughly rinsed with deionized water and ultrasonically cleaned in pure ethanol.Simultaneous acquisition of surface topography and Volta potential difference was conducted using a SCM-PIT probe working in a tapping/lift mode sequence with an interleave lift height of 100μm.All scans were conducted in ambient air at 25°C and～40% relative humidity using a scanning frequency of 0.5Hz,a pixel resolution of 512×512,and a zero bias voltage.Results were analyzed by using NanoScope Analysis 1.5 software,in which topography maps were flattened using 1st order flattening and the Volta potential maps were plane fitted against theα-Mg matrix to calculate the relative nobility of theβ-Mg17Al12precipitates against the surroundingα-Mg matrix.For convention,large Volta potential differences(light regions)correspond to higher absolute work function values between the microstructural feature and the AFM tip,and hence,higher electronic activity(net anodic behavior).According to this definition,lower Volta potential differences(dark regions)indicate lower electronic activity(net cathodic behavior).

2.4.Electrochemical and corrosion measurements

The electrochemical response and corrosion behavior of the ECAPed alloys with different levels of deformation strain were investigated by electrochemical microcell technique,immersion testing,weight loss measurements,and global electrochemical measurements.All electrochemical and corrosion measurements of the ECAPed specimens were performed on the flow plane.Samples of 1.25cm in diameter and 2mm in thickness were used for the electrochemical and corrosion testing.Prior to each test,samples were mechanically ground with SiC abrasive paper up to 1200 grit and then polished with diamond and silica colloidal suspensions of up to 0.05μm particle size.After polishing,samples were degreased with ethanol in an ultrasonic bath for 5min and dried with compressed air.

2.4.1.Electrochemical microcell technique

The electrochemical microcell technique was used to investigate the local electrochemical behavior of the recrystallized and un-recrystallized regions of the ECAPed alloy after four passes exposed to a 3.5wt% NaCl solution.The entire setup is mounted on a microscope allowing for positioning of the microcell on the specimen surface.The microcell consists of a glass capillary with an opened mouth diameter of 70μm.The front end of the microcell is adhered to the site of interest with a layer of silicone rubber.The capillary is filled with 3.5wt%NaCl electrolyte.The microcell is fixed at the microscope carousel,replacing an objective,and the specimen is mounted on the microscope stage.This setup facilitates the search of recrystallized and un-recrystallized regions in the ECAPed alloy using the microscope before switching to the microcell.A saturated calomel electrode(SCE)and a platinum wire were used as the reference and counter electrodes,respectively.Potentiodynamic polarization measurements in the recrystallized and un-recrystallized regions were performed from?250mV to+500mV vs.OCP or until a current density of 1mA cm?2was reached at a scan rate of 5mV/s using a Gamry,Interface 1000TMPotentiostat/Galvanostat/ZRA.The corroded morphology of the tested sites was further examined using SEM.

2.4.2.Immersion testing

The immersion testing of the ECAPed alloys was performed in a 3.5wt% NaCl solution at room temperature for 1h and 24h to investigate the corrosion initiation and propagation,respectively.The solution was not renewed during the immersion period and the samples were suspended from a nylon string in 800ml of solution without agitation.Optical microscope images were taken before and after immersion for 1h to clearly identify the preferential corrosion initiation sites.After the immersion testing,samples were chemically cleaned for 1min with a pickling solution containing 200g/L CrO3and 10g/L AgNO3.After removal of corrosion products,specimens were rinsed with deionized water and ethanol and dry with compressed air.Morphology of the corroded specimens after removal of corrosion products were characterized by SEM,in addition,surface topography maps and roughness profiles were obtained from a 3D profilometer(Keyence VHX-6000).

2.4.3.Weight loss measurements

The corrosion rate of the ECAPed alloys after different ECAP passes was calculated from weight loss measurements.ECAPed specimens were immersed for seven days in a 3.5wt% NaCl solution at room temperature.Similar setup as the immersion testing was used for the weight loss measurements.After the weight loss test,samples were also chemically cleaned in the pickling solution as described above.Then,the samples were rinsed with deionized water and dried by cold air flow.The dried specimens were weighed before and after the immersion test using a digital balance with a precision of 0.1mg.The weight loss test was repeated three times to confirm the reproducibility of the results and the average values were reported.The corrosion rate(in mmpy)from the weight loss measurements was calculated from the following equation[19]:

where(mg)is the weight loss,A(cm2)is the exposed surface area of the specimens,andt(day)is the immersion time.

2.4.4.Electrochemical measurements

Open circuit potential(OCP),electrochemical impedance spectroscopy(EIS),and potentiodynamic polarization(PDP)measurements were performed on the ECAPed samples in a 0.05M NaCl solution at room temperature.The diluted NaCl solution was used to guarantee passivation conditions allowing to analyze the influence of strain level on the protective properties of the passive film.Electrochemical measurements were acquired using a Gamry,Interface 1000TMPotentiostat/Galvanostat/ZRA and a flat cell with an exposed area of 1 cm2in a conventional three-electrode cell configuration with 300mL of electrolyte.A saturated calomel electrode(SCE)and a platinum mesh were used as the reference electrode and counter electrode,respectively.OCP was measured for 1 h to achieve steady state conditions.Then,EIS measurements were carried out at OCP using a sinusoidal potential signal with an AC amplitude of 10mV over a frequency range from 100kHz to 10 mHz with 10 points per decade.The EIS data were fitted with equivalent circuits using the EC-lab V10.40 software.Finally,PDP tests were performed from?250mV vs.OCP to 500mV vs.OCP or until a current density of 100μA cm?2was achieved using a scan rate of 1mV s?1.Corrosion current density(icorr)was determined by extrapolating the cathodic Tafel slope to the corrosion potential(Ecorr).The breakdown potential(Eb)was also obtained from the PDP curves.All electrochemical measurements were performed at least three times to guarantee repeatability.

3.Results and discussion

3.1.Microstructural characterization

3.1.1.Initial microstructure prior to ECAP

Fig.2 shows the microstructure of the as-received cast alloy(Fig.2(a)),the homogenized alloy at 413°C for 24h(Fig.2(b)),and the alloy after pre-heating at 310°C for 30min before ECAP processing(Fig.2(c)and(d)).As seen in Fig.2(a),the as-cast alloy exhibits a typical dendritic microstructure with coarse grains of approximately 375±145μm.This dendritic microstructure consists of a primaryα-Mg matrix,lamellarβ-Mg17Al12precipitates,a partially divorced eutecticα+βphase,and Al8Mn5intermetallic particles.After homogenization treatment at 413°C for 24h,most of the lamellar and eutectic phases are dissolved into theα-Mg matrix(Fig.2(b)),thus resulting in a supersaturated solid solution with an average grain size of 515±99.5μm.Fig.2(b)also shows that the Al8Mn5intermetallic particles are still present after solution heat treatment,owing to the low solubility of Mn in the Mg matrix as well as their high thermal stability[20,21].Typically,some residual eutecticβ-Mg17Al12phase that is not dissolved during homogenization might still be present.During pre-heating to the ECAP temperature of 310°C for 30min before the ECAP processing,large amounts ofβ-Mg17Al12precipitates form at the grain boundaries as well as in the grain interiors(Fig.2(c)).The high magnification SEM image in Fig.2(d)shows the formation of lamellarβ-Mg17Al12precipitates at the grain boundaries(known as discontinuousβ-Mg17Al12precipitates)and lath-shapedβ-Mg17Al12precipitates within theα-Mg grains(known as continuousβ-Mg17Al12precipitates).Similar microstructures with lamellar discontinuousβ-Mg17Al12precipitates at the grain boundaries and lath-shaped continuousβ-Mg17Al12precipitates at the grain interior have been reported by Xu et al.[22]during pre-heating of a homogenized AZ91 alloy at 350°C for 5min before hot compression.The discontinuous precipitation is characterized by cellular growth of alternating layers ofβ-Mg17Al12phase and a less saturatedα-Mg phase at high angle grain boundaries,thus resulting in a lamellar structure that grows behind a moving grain boundary into theα-Mg matrix[22,23].In contrast,during continuous precipitation,lath-shapedβ-Mg17Al12precipitates nucleate and grow inside the originalα-Mg grains,preferentially at defect sites such as vacancies and dislocations[23].

3.1.2.Microstructure evolution during ECAP processing

Fig.3 shows optical micrographs of the ECAPed alloys after multiple ECAP passes(N=1–4).After the first ECAP pass,the ECAPed alloy exhibits a bimodal grain structure consisting of original coarse grains that are elongated in the shearing direction and are surrounded by new recrystallized fine grains.As seen in Fig.3(a),the coarse grains occupy a significantly larger volume fraction than the fine grains.This bimodal structure and the presence of equiaxed fine grains along the original grain boundaries indicates that DRX occurred during the ECAP process[7,22].Increasing the strain level(i.e.,the number of ECAP passes)results in a gradual increase in the volume fraction of fine grains,also known as DRX ratio[24],which increased from 39% after the first pass up to 77% after the fourth pass.Nevertheless,a fully recrystallized microstructure was not achieved,and coarse grains still existed even after four passes.In this study,the microstructure development of the AZ91 alloy after processing by ECAP is consistent with the model proposed by Figueiredo and Langdon[25,26],who have suggested that for magnesium alloys with a coarse-grained initial microstructure,a bimodal grain microstructure with the fine grains forming a necklace-like structure along the original grain boundaries is obtained after the first ECAP pass.They have also proposed that this bimodal microstructure becomes more homogeneous as the number of passes increases until full recrystallization.Of note,the presence of a bimodal microstructure even after four ECAP passes is associated not only with the initial coarse-grained structure of the homogenized AZ91 alloy but also with the formation of fineβ-precipitates during ECAP processing,which can retard the grain growth of DRXed grains through a pinning effect that limits the grain boundary motion[27].Consequently,the original coarse grains can still occupy a relatively high volume fraction even after several ECAP passes.Therefore,the development of a fully recrystallized microstructure in AZ91 alloys might require more passes than predicted by the model proposed by Figueiredo and Langdon.

Fig.2.SEM micrographs of the microstructure of(a)as-cast alloy,(b)homogenized alloy and(c and d)alloy after pre-heating at 310°C for 30min before ECAP processing.

Fig.3.Optical micrographs of the ECAPed specimens with various number of ECAP passes(a)N=1(b)N=2,(c)N=3 and(d)N=4.

Fig.4.SEM micrographs of the ECAPed alloy after the first pass(N=1)showing(a,b)bimodal grain structure with bulging of the initial grain boundaries and presence of large amount ofβ-Mg17Al12 precipitates in both the un-DRXed and DRXed regions and(c,d)DRX occurring in neighboring regions of an undissolved eutecticβ-phase demonstrating the occurrence of PSN mechanism.

SEM micrographs of the ECAPed alloy after the first ECAP pass are shown in Fig.4.As shown in Fig.4(a),the microstructure after the first ECAP pass exhibits the aforementioned bimodal grain structure consisting of the original elongated coarse grains and the newly formed equiaxed fine grains,denoted in Fig.4(a)as un-DRXed and DRXed regions,respectively.The average grain size of the DRXed grains is 4.08±0.75μm.DRX in AZ91 alloys usually occurs in a discontinuous form(i.e.,through discontinuous DRX),in which nucleation and subsequent growth of DRXed grains occurs along the original grain boundaries,where the stress concentration is higher,and thus both basal and non-basal slip systems are more easily activated,thereby resulting in the formation of a necklace-type structure[7,28,29].Fig.4(b),corresponding to the enlarged view of the micrograph in Fig.4(a),shows that recrystallization of new fine grains is accompanied by bulging of the original grain boundaries,with the new grains growing into the primary coarse grains(as indicated by the red arrows in Fig.4(b)),a finding indicative of strain-induced grain boundary migration and therefore the occurrence of discontinuous DRX[21,30].Fig.4(b)also shows the presence of large amounts ofβ-Mg17Al12precipitates in both the un-DRXed and DRXed regions.β-Mg17Al12precipitates with a relatively spherical morphology are seen along the grain boundaries as well as in the grain interior of the DRXed grains.In contrast,lath-shaped continuous precipitates are seen at the grain interior of the original coarse grains.This microstructure is also consistent with the microstructure reported by Xu et al.[22]during hot compression of a homogenized AZ91 alloy.The spherical shape of the precipitates formed in the DRXed regions was found to be more energetically favorable than the lamellar discontinuous precipitates and lath-shaped continuous precipitates formed during traditional heat treatment,owing to the high density of defects introduced during the ECAP process[6,7].The abundantβ-Mg17Al12precipitates are attributable to the formation of precipitates during the pre-heating process,as shown in Fig.2(c)and(d),fracturing of pre-existing precipitates by shearing,and strain-induced dynamic precipitation(SIDP)from the supersaturatedα-Mg solid solution during the ECAP process[7,31,32].Dynamic precipitation at the DRXed grain boundaries(GBs)is promoted,because the grain boundaries provide rapid diffusion paths for the solute atoms.In addition,the dynamic precipitation observed at the grain interior of the un-DRXed and DRXed grains can be attributed to the high density of defects,such as dislocations and vacancies,which can serve as heterogeneous nucleation sites forβ-Mg17Al12precipitates.These defects also accelerate the diffusion of solute atoms through theα-Mg matrix[23,29].The average size of theβprecipitates at the DRXed GBs is 1.46±0.43μm,whereas theβprecipitates at the interior of the DRXed grains are smaller than 100nm.DRX is greatly influenced by the size of the precipitates,in which coarseβprecipitates(diameter>1μm)promote dynamic recrystallization through the particle-stimulated nucleation(PSN)mechanism,whereas smaller precipitates(diameter≤1μm)can have a grain boundary pinning effect that restricts the growth of the recrystallized grains and promotes grain refinement[28,30,33,34].In the PSN mechanism,the nucleation of DRXed grains is facilitated by the high stored strain energy at the precipitate/matrix interface that occurs during the severe deformation and leads to accumulation of dislocations in the vicinity of coarse precipitates[32,35].These regions with a high density of dislocations serve as nucleation sites for new DRXed grains[35].Fig.4(c)and(d)shows the presence of an un-dissolved eutecticβ-phase surrounded by DRXed grains,thus providing evidence that largeβ-Mg17Al12particles can act as nucleation sites for DRX through the PSN mechanism.

Fig.5.SEM micrographs of the ECAPed specimens with various number of ECAP passes showing(a–d)the bimodal grain structure,(e–h)a DRXed region and(i–l)an un-DRXed region after N=1,N=2,N=3,and N=4,respectively.The values in(e–h)indicate the average grain size of the DRXed grains.

Fig.5 shows the microstructure evolution of the ECAPed specimens as increasing the level of deformation strain.As shown in Fig.5(a–d),a bimodal grain structure still exists even after four ECAP passes.However,the DRXed ratio increases with the number of ECAP passes,thus resulting in a more uniform recrystallized microstructure.These values are reported in Table 1.The DRXed regions after different ECAP passes are shown in Fig.5(e–h).These micrographs show that the average grain size of the DRXed grains decreases from 4.08±0.75μm after one pass to 1.58±0.40μm after four passes.This substantial grain refinement may be associated with the combined effects of DRX and SIDP during ECAP processing[29],in which the severe imposed strain promotes dynamic recrystallization of new fine grains,and theβprecipitates can either accelerate DRX through the PSN mechanism or restrict grain growth of the DRXed grains through a grain boundary pinning effect.The concurrent presence of dynamic precipitation and recrystallization during ECAP processing has been denoted as the combined-reaction region,which is characterized by fine grains formed at the original grain boundaries and the presence of submicronβprecipitates[29].Fig.5(e–h)also shows that the size and volume fraction ofβprecipitates in the DRXed regions decrease and increase,respectively,with increasing strain level.These values are also reported in Table 1,which shows that the volume fraction of precipitates increases from 8.57%after one pass to 20.65% after four passes,whereas the average size decreases from 1.86±0.43 after the first pass to 0.73±0.34 after the fourth pass.Of note,by decreasing the deformation temperature after each ECAP pass,grain growth is prevented not only by lower mobility of grain boundaries at lower temperatures but also by the greater amount of precipitates formed at lower temperatures,which facilitate recrystallization and retard the growth of the DRXed grains[7,9].Increasing the temperature is well known to result in dislocation annihilation by cross-slip;therefore,by decreasing the temperature,there is a higher density of dislocations that can serve as nucleation sites for dynamic precipitation[7].In addition,the presence of finer precipitates as increasing the number of ECAP passes suggests that the pinning effect becomes more predominant as the level of deformation strain increases[27,36].Fig.5(i–l)shows the evolution of the un-DRXed regions as the number of ECAP passes is increased,in which it can be noticed that a large volume fraction of lath-shaped continuousβprecipitates is formed with increasing strain level(from 16.32% after the first pass to 32.66% after the fourth pass;Table 1).

Fig.6 shows the XRD patterns of the AZ91 alloy after each ECAP pass.As shown in Fig.6(a),the ECAPed alloys are mainly composed ofα-Mg andβ-Mg17Al12phases.Fig.6(a)also shows that the intensity of the peaks corresponding to the basal planes(0002)and prismatic planes(01)varies as the number of ECAP passes increases,and notably,the prismatic texture is stronger with increasing numbers of passes.This behavior suggests that the texture evolves during severe deformation[6,7,29].Fig.6(b)shows an enlarged view of the three major peaks of the Mg matrix,in which some of these peaks(highlighted by the vertical dashed lines)split and shift to lower angles as the number of ECAP passes increases.Ma et al.[29]have attributed this behavior to different Al content between the original un-DRXed grains and the newly formed DRXed grains.

Table 1Microstructural parameters of AZ91 alloys after different ECAP passes.

Fig.6.(a)XRD patters of the ECAPed specimens with various number of ECAP passes.(b)Enlarged patterns for the three major peaks of the Mg matrix showing splitting and shifting of the peaks as increasing the number of ECAP passes.

3.2.Electrochemical and corrosion measurements

3.2.1.Corrosion initiation in ECAPed alloys with bimodal grain structure

The corrosion initiation of ECAPed AZ91 alloys at different levels of deformation strain was investigated after immersion in 3.5wt% NaCl solution for 1h.Fig.7 shows optical images of the ECAPed alloys withN=1(a and b)andN=4(d and e)before and after immersion in the electrolyte solution.The micrographs of the corroded specimens were taken after the removal of corrosion products.These images show that corrosion initiates in the DRXed regions,but there is no evidence of corrosion in the un-DRXed grains.To provide further insight into the morphology of the corrosion initiation in the DRXed regions,Fig.7(c)and(f)shows SEM micrographs of the ECAPed alloys after the short immersion test.Corrosion clearly initiates at the DRXed regions and appears to occur at the DRXed grain boundaries and to subsequently extend into the interior of the grains.This corrosion initiation at the DRXed grain boundaries occurs as a consequence of microgalvanic coupling between theβ-Mg17Al12precipitates and the adjacentα-Mg matrix,thus leading to preferential dissolution of theα-Mg matrix and undermining theβ-phase,which eventually detaches from the alloy.According to these results,theβ-Mg17Al12phase formed at the DRXed grain boundaries acts as an effective galvanic cathode that accelerates the anodic dissolution of theα-Mg matrix.This result is in agreement with the dual role of theβ-Mg17Al12phase on the corrosion behavior of Mg–Al alloys,in which the presence of discontinuousβ-Mg17Al12precipitates along the grain boundaries induces galvanic corrosion as these precipitates act as strong galvanic cathodes[37].Comparison of Fig.7(c)and(f)suggests that corrosion in the ECAPed alloy after the first pass is more severe and exhibits a greater penetration depth than the corrosion observed in the ECAPed alloy after the fourth pass.This behavior may be associated with the larger size of theβ-Mg17Al12precipitates formed at the DRXed grain boundaries of the ECAPed alloy after the first pass,as shown in Table 1,thus potentially leading to deeper corrosion pits after the precipitates are peeled off from the alloy surface.In contrast,the smaller size of theβ-Mg17Al12precipitates in the ECAPed alloy after the fourth pass results in shallow corrosion pits.In addition to the intergranular corrosion occurring at the DRXed grain boundaries,corrosion within the DRXed grains also occurs,as highlighted by the enclosed yellow regions in Fig.7(f).This behavior is associated with micro-galvanic coupling between the nano-sized sphericalβ-Mg17Al12precipitates formed at the interior of the DRXed grains and the adjacentα-Mg matrix.It is worth to notice that this intragranular corrosion process is less aggressive than the intergranular corrosion occurring at the DRXed grain boundaries,mainly because of the very low cathodeto-anode area ratio between the nano-sized precipitates at the interior of the grains and the neighboringα-Mg matrix,which leads to slow galvanic corrosion kinetics.

Fig.7.Optical micrographs of the ECAPed alloys with(a,b)N=1 and(d,e)N=4 before and after immersion for 1h in 3.5wt% NaCl solution.High magnification SEM micrographs of the(c)ECAPed alloy with N=1 and(f)ECAPed alloy with N=4 after immersion for 1h in 3.5wt% NaCl solution showing corrosion initiation in the DRXed regions.

To investigate the causes of the preferential corrosion of the DRXed regions,we performed elemental mapping by STEM/EDS on the recrystallized and un-recrystallized regions of an ECAPed alloy,and examined differences in elemental composition between theβ-Mg17Al12and theα-Mg phases present in each region.Fig.8 shows STEM micrographs,EDS maps,and point analysis from an un-DRXed and a DRXed region of an ECAPed alloy after four passes.The micrograph in Fig.8(a)corresponds to the un-DRXed region,showing the presence of the lath-shaped continuousβ-Mg17Al12precipitates,whereas Fig.8(e)corresponds to the DRXed region,showing the presence ofβ-Mg17Al12precipitates formed at the DRXed grain boundaries and at the grain interior.From the EDS maps of Mg(Fig.8(b)and(f)),Al(Fig.8(c)and(g))and Zn(Fig.8(d)and(h)),a higher chemical difference is observed between the matrix and precipitates in the DRXed region than the un-DRXed region.This finding was further confirmed by elemental point analysis performed on theβ-Mg17Al12phase andα-Mg matrix in the DRXed and un-DRXed regions.From EDS analysis,it was found that theβ-Mg17Al12precipitates formed in the DRXed grains have greater aluminum content(36.2 at%)than the lath-shaped continuousβ-Mg17Al12precipitates in the un-DRXed grains(33.6 at%).In contrast,the Al content in the DRXed grains is lower(3.06 at%,measured at points 1–3 in Fig.8(e))than that in the un-DRXed grains(4.05 at%,measured at points 1–3 in Fig.8(a))as noticed in Fig.8(i).The greater microchemistry difference between theβ-Mg17Al12precipitates andα-Mg matrix in the DRXed regions provides a greater driving force for microgalvanic coupling,as compared with the galvanic corrosion activity developed in the un-DRXed region,thus leading to preferential corrosion of the DRXed regions upon immersion.Ma et al.[29]have reported similar findings when investigating the mechanisms of dynamic precipitation and recrystallization of a Mg–9wt%Al alloy processed by ECAP.They reported that the recrystallizedα-Mg grains exhibit lower Al solute content than the heavily deformed coarse grains.They have attributed these findings to the combined actions of dynamic precipitation and recrystallization.Such process promotes the formation of recrystallized grains with a near equilibrium concentration of Al,which is lower than the excess Al solute content in the original supersaturatedα-Mg matrix.Indeed,they proposed that the decrease in chemical potential energy due to the lower Al content in the recrystallized grains is the dominant thermodynamic driving force for further recrystallization and the formation of sub-micronα-Mg grains.

Of note,dislocation density can also play a role in the corrosion behavior of severely deformed magnesium alloys.Typically,regions with a high density of dislocations results in higher anodic dissolution rates,owing to the higher distorted energy of the crystallographic defects,thus making these sites more electrochemically active for corrosion processes[3,13,38].In partially recrystallized microstructures,the recrystallized grains are well known to exhibit relatively lower dislocation density than the heavily deformed un-recrystallized grains,which have a significantly higher dislocation density[6,7].In that sense,the un-recrystallized grains would be expected to be more susceptible to corrosion than the recrystallized grains.However,this study shows that the preferential corrosion initiation and propagation occurs in the recrystallized grains,although they might exhibit a lower dislocation density than the un-recrystallized grains.This behavior indicates that,in the ECAPed alloys investigated in this study,the greater difference in microchemistry between theβ-Mg17Al12precipitates and theα-Mg matrix in the DRXed regions plays a more crucial role in the corrosion initiation and propagation than the presumably higher dislocation density in the un-DRXed grains.This corrosion behavior is opposite from that of partially recrystallized aluminum alloys,in which corrosion preferentially occurs in the un-recrystallized grains because of the high level of stored energy(i.e.,high dislocation density)of the subgrain boundaries in these grains,which predominates over the intergranular corrosion at the grain boundaries decorated with secondary phases but with less stored energy[39,40].

Fig.8.STEM micrographs of the ECAPed alloy with N=4 showing(a)lath-shapedβ-Mg17Al12 precipitates formed in the un-DRXed regions and(e)sphericalβ-Mg17Al12 precipitates formed in the DRXed regions with their corresponding EDS maps of(b,f)Mg,(c,g)Al and(d,h)Zn,respectively.(i)Average elemental composition of theα-Mg matrix in the DRXed and un-DRXed regions.

The out-of-plane crystal orientation of an ECAPed alloy with four passes was examined by electron backscatter diffraction to analyze the crystallographic orientation of the DRXed and un-DRXed regions.The inverse pole figure map shown in Fig.9(a)is divided into fine grain and coarse grain inverse pole figure maps on the basis of grain size(6μm).As shown in Fig.9(a),dense fine grains occupy most of the region after four ECAP passes,thus indicating a high DRX ratio,and the original elongated coarse grains are surrounded and intersected by these newly formed fine grains.The pole figure of the(0002)plane orientation and maximum intensity of the un-DRXed region(max.intensity=51.17)indicate a stronger basal texture than the DRXed region(max.intensity=5.77).These results are in good agreement with findings from previous studies investigating the crystal orientation of partially recrystallized magnesium alloys[8,27,41].To further display the DRXed region,Fig.9(b)shows a zoomin of the black rectangle in Fig.9(a).From Fig.9(b),the average grain size of the equiaxed DRXed grains is 1.82μm±1.96,which is similar to the values reported in Table 1.The(0002)pole figure in Fig.9(b)exhibits a similar texture to that of the DRXed grains in Fig.9(a),thus confirming the weak basal texture of the DRXed region compared with the un-DRXed region.These results clearly reveal that the un-DRXed grains preferentially exhibit a basal texture,whereas the DRXed grains are characterized by a more random texture.As mentioned above,crystallographic orientation can also play an important role in the corrosion behavior of magnesium alloys.Grains with basal crystal orientation have been widely and consistently reported to be more corrosion resistant than grains with a non-basal crystal orientation[4,8,15–17].The greater corrosion resistance of the closely packed basal plane has been associated with its higher atomic density,which results in higher binding energy,lower surface energy,and higher activation energy for removal of atoms from the metal lattice[11,17].Thus,the basal plane exhibits greater electrochemical stability and corrosion resistance than the less packed non-basal planes.Therefore,the DRXed region with a weak basal texture in the ECAPed alloy with four passes is expected to be more prone to corrosion than the highly textured un-DRXed region.Thus,in addition to the greater microchemistry difference between theβ-Mg17Al12phase andα-Mg matrix in the DRXed region,the more random crystal orientation of the DRXed grains can also explain why corrosion preferentially initiates in the DRXed region.These results are in agreement with those from a previous study by Luo et al.[8],who have investigated the effect of grain size and crystal orientation on the corrosion behavior of an extruded Mg–6Gd–2Y–0.2Zr alloy.They have found that small grains with non-basal orientation are the preferential sites of corrosion initiation in a 5wt% NaCl solution.According to the results reported to date,the combined effect of lower microchemistry difference between precipitate and matrix and strong basal texture in the un-DRXed grains is believed to predominate over the presumably higher dislocation density in these grains,such that they are more corrosion resistant than the DRXed grains.

Fig.9.EBSD IPF maps(out-of-plane crystal orientation)of ECAPed alloy with N=4:(a)large region mapping with both DRXed grains and un-DRXed grains with their corresponding(0002)pole figures and(b)zoom-in map from the black rectangle in(a)showing detailed grains information of the DRXed region along with the corresponding(0002)pole figure.

3.2.2.Localized Volta potential distribution in the recrystallized and non-recrystallized regions

SKPFM measurements were performed to further support the above results regarding the stronger microgalvanic coupling in the DRXed regions,thus leading to preferential corrosion initiation in these regions.SKPFM enables investigation of the local nobility of the different constituents in a material and a prediction of potential cathodes and anodes that will be developed during immersion testing in corrosion environments.Fig.10 shows the topography(a)and(d)and Volta potential maps(b)and(e)from the un-DRXed and DRXed regions of an ECAPed alloy after four passes.The Volta potential is associated with the electronic activity of a specific phase in the material;typically,the larger the Volta potential with respect to the probe,the greater the electrochemical activity under immersion conditions[42].In that sense,dark and light areas in the Volta potential maps correspond to cathodic and anodic sites,respectively.Fig.10(b)and(e)shows that theβ-Mg17Al12precipitates formed in both the un-DRXed and DRXed regions are cathodic against theα-Mg matrix,and there is a large potential difference between them.This cathodic behavior is expected,because theβ-phase is more enriched in Al than the Al-depletedα-Mg matrix.According to these results,these precipitates are expected to induce microgalvanic corrosion at theβ-phase/α-Mg matrix interphase when the ECAPed alloy is immersed in an aggressive solution.However,the line profiles in Fig.10(c)and(f)show that the absolute value of the Volta potential difference between theβ-Mg17Al12precipitates formed in the DRXed grain boundaries and theα-Mg matrix is higher(86mV)than that between the lath-shaped continuousβ-Mg17Al12precipitates and theα-Mg matrix in the un-DRXed region(73mV).These results indicate that theβ-phase andα-Mg matrix in the DRXed region form a stronger galvanic coupling,and therefore the DRXed regions are more susceptible to microgalvanic corrosion.These results support the difference in chemical composition observed from the STEM/EDS analysis shown in Fig.8,in which there is a greater difference in microchemistry between theβ-phase and theα-Mg matrix in the DRXed region than the un-DRXed region,thus providing a greater driving force for microgalvanic corrosion.

3.2.3.Localized electrochemical behavior of recrystallized and non-recrystallized regions

The local electrochemical behavior of the DRXed and un-DRXed regions of an ECAPed alloy with four ECAP passes was investigated with the electrochemical microcell technique.Fig.11(a)and(b)shows optical images of the ECAPed alloy with four passes,indicating the DRXed and un-DRXed regions before and after the potentiodynamic polarization test in 3.5wt% NaCl solution.The local potentiodynamic polarization measurements in the DRXed and un-DRXed regions are shown in Fig.11(c).From these PDP curves,the un-DRXed region clearly exhibits a more noble corrosion potential(?1.495V vs.SCE)than that in the DRXed region(?1.564V vs.SCE).Furthermore,the corrosion current density of the un-DRXed region(4.16μA/cm2)is lower than that of the DRXed region(10.2μA/cm2).The lower electrochemical activity of the un-DRXed grains is consistent with the SKPFM results showing a smaller Volta potential difference between the lath-shaped continuousβ-Mg17Al12precipitates andα-Mg matrix in the un-DRXed grains than that in the DRXed regions between theβ-Mg17Al12precipitates at the DRXed grain boundaries and the neighboringα-Mg matrix.These local electrochemical results confirmed that the DRXed region exhibits faster corrosion reaction kinetics than the un-DRXed region,in agreement with the preferential corrosion initiation in the DRXed grains observed after the short-term immersion testing shown in Fig.7.In addition,the faster corrosion kinetics of the DRXed region suggests that corrosion propagation under longer immersion times tends to proceed along the DRXed grains,and less corrosion attack should be expected in the un-DRXed grains.As explained above,the higher electrochemical activity of the DRXed regions is associated with the greater microchemistry difference between theβ-Mg17Al12precipitates formed at the DRXed grain boundaries and the adjacentα-Mg matrix,which induces stronger microgalvanic coupling.The higher corrosion susceptibility of the DRXed grains is also likely to be associated with the weaker basal texture of these grains compared to the highly textured un-DRXed grains,thus making the DRXed grains more vulnerable to electrochemical reactions.Fig.11(d)and(f)shows SEM micrographs of the un-DRXed and DRXed regions,respectively,after local potentiodynamic polarization tests and removal of the corrosion products.Comparison of Fig.11(d)and(f)indicates that corrosion appears to be more aggressive in the DRXed region than in the un-DRXed region,in accordance with the higher corrosion current density obtained for the DRXed region.Furthermore,Fig.11(f)clearly shows that corrosion in the DRXed region occurs mainly at the DRXed grain boundaries,owing to microgalvanic coupling between theβ-Mg17Al12precipitates formed at the grain boundaries and the adjacentα-Mg matrix,which leads to preferential corrosion of the matrix.Minor corrosion was also observed at the interior of the DRXed grains(highlighted by the yellow circles in Fig.11(f)),which,as mentioned above,is associated with micro-galvanic coupling between the nanosizedβ-Mg17Al12precipitates within the DRXed grains and the adjacentα-Mg matrix.This micro-galvanic coupling is not likely to play a significant role in the corrosion propagation of the DRXed region,owing to the significantly small cathode-to-anode area ratio,which drastically limits the reaction kinetics of this galvanic process.Micro-galvanic corrosion is also observed in the un-DRXed region(Fig.11(d))occurring at the interface between the lath-shaped continuousβ-Mg17Al12precipitates and the surroundingα-Mg matrix.However,corrosion also occurs in the bulk of theα-Mg matrix,as highlighted by the yellow circles in Fig.11(d).These results indicate that the microgalvanic coupling between the precipitate and matrix in the un-DRXed region is relatively weak,such that microgalvanic corrosion is not the only corrosion mechanism occurring in the un-DRXed region,and instead localized corrosion in the bulk matrix can also occur through preferential anodic dissolution.

Fig.10.SKPFM measurements of the ECAPed alloy with N=4 showing(a,d)topography,(b,e)Volta potential difference map,and(c,f)the corresponding line profile along the white arrow of a lath-shapedβ-Mg17Al12 precipitate in the un-DRXed grains and a sphericalβ-Mg17Al12 precipitate in the DRXed grains,respectively.

Fig.11.Optical micrographs of the ECAPed alloy with N=4(a)before and(b)after the local potentiodynamic polarization test in 3.5wt.% NaCl using the electrochemical microcell technique.(c)Local potentiodynamic polarization curves of the DRXed and un-DRXed regions in the ECAPed alloy with N=4.SEM micrographs of the(d)un-DRXed region and(f)the DRXed region after the local electrochemical test and after removal of corrosion products.

3.2.4.Corrosion propagation in ECAPed alloys with bimodal grain structure

Immersion testing for longer times was also performed to investigate the corrosion propagation in ECAPed alloys at different levels of deformation strain.Fig.12 shows SEM micrographs of the ECAPed alloys withN=1 andN=4 after immersion in a 3.5wt% NaCl solution for 24h and after removal of corrosion products.Fig.12(a-c)shows the corroded morphology of the ECAPed alloy after the first pass,and Fig.12(d–f)corresponds to the morphology of the ECAPed alloy after the fourth pass.Comparison of the low magnification SEM micrographs in Fig.12(a)and(d)clearly shows that the ECAPed alloy after the first pass exhibits a more severe and localized corrosion than the ECAPed alloy after the fourth pass,which shows a more uniform corrosion morphology with a relatively shallow penetration depth.As mentioned above,the deeper penetration observed for the ECAPed alloy after the first pass is associated with the larger size of theβ-Mg17Al12precipitates along the DRXed grain boundaries that after peeling off from the alloy surface results in wider and deeper corrosion pits.The larger corrosion pits formed on the ECAPed alloy after the first pass expose a larger active area that further accelerates the corrosion rate at these regions.In addition,corrosion preferentially initiates in the DRXed regions,owing to the stronger micro-galvanic coupling between theβ-Mg17Al12precipitates along the DRXed grain boundaries and the adjacentα-Mg matrix.According to this finding and the low DRX ratio of the ECAPed alloy ratio after the first pass(i.e.,lower volume fraction of DRXed grains;39% forN=1),corrosion in this alloy is expected to appear more localized and to penetrate deeper into the DRXed regions of the material.In contrast,the smaller size of theβ-Mg17Al12precipitates along the DRXed grain boundaries of the ECAPed alloy after the fourth pass combined with a higher DRXed ratio results in a more uniform corrosion that spreads through the high volume fraction of DRXed grains.The higher volume fraction of well-dispersed fineβ-Mg17Al12precipitates formed in the ECAPed alloy after the fourth pass is also likely to further promote a more uniform corrosion morphology,due to the numerous microgalvanic cells form over almost the entire alloy surface that prevents the development of a highly localized attack.According to high magnification SEM images,corrosion in both ECAPed alloys is mainly concentrated in the DRXed regions(Fig.12(c)and(f)).For the ECAPed alloy after the first pass,severe corrosion is observed in the DRXed regions,as shown in Fig.12(c),in which corrosion initiates along the grain boundaries because of microgalvanic corrosion,and it propagates into the grain interiors.Fig.12(c)also reveals that in addition to the microgalvanic corrosion occurring at the DRXed grain boundaries,corrosion in the un-DRXed regions is also evident,in which corrosion appears to be initiated at the interphase between the lath-shaped precipitates and the adjacentα-Mg matrix because of microgalvanic corrosion between these phases.However,the corrosion in the un-DRXed region is significantly less severe than that in the DRXed region.Fig.12(f)shows that the corrosion in the ECAPed alloy after the fourth pass is markedly less aggressive than that in the ECAPed alloy after the first pass.Furthermore,corrosion mainly occurs in the DRXed region,and almost no corrosion attack is identified in the un-DRXed region.This behavior is associated with microstructural changes in the recrystallized and un-recrystallized regions,in which increasing the level of deformation strain induces a more distinct difference in the electrochemical activity of the DRXed and un-DRXed regions,such that corrosion preferentially initiates and propagates through the DRXed regions.These observations are in agreement with the local PDP results obtained from the electrochemical microcell technique,in which a higher corrosion current density is observed in the DRXed region than the un-DRXed region,thus indicating that the DRXed regions indeed have faster corrosion kinetics that promotes corrosion development along these regions.

Fig.12.SEM micrographs of the ECAPed alloys with(a,b,c)N=1 and(d,e,f)N=4 after immersion for 24h in 3.5wt% NaCl solution.

To evaluate the penetration depth of the ECAPed alloys at different levels of deformation strain,3D profilometry studies were performed after immersion for 24h.Fig.13 shows topography and roughness profiles of the ECAPed alloys withN=1 andN=4 after immersion for 24h and after removal of corrosion products.In accordance with results in Fig.12,the topography maps show that corrosion in the ECAPed alloy after the first pass is more localized and exhibits a greater penetration depth than the ECAPed after the fourth pass,which is characterized by a more uniform corrosion with a shallow penetration depth.Fig.13 also includes linear roughness profiles along the regions with the highest penetration(i.e.,blue regions),to analyze the changes in penetration depth with increasing strain.The roughness parameters,Ra(arithmetical mean height)and Rz(maximum height profile)were extracted from the depth profile;these parameters represent the average roughness and the maximum roughness along the line profile,respectively.The uncorroded areas served as the baseline for calculating the maximum roughness,such that these values correspond to the maximum pit depth.In that sense,the maximum pit depth in the ECAPed alloy after the first pass is substantially higher(113.64μm)than the maximum pit depth in the ECAPed alloy after the fourth pass(7.71μm).These results confirm that smaller DRX ratios combined with larger precipitates along the DRXed grain boundaries promote the development of a localized corrosion morphology with high penetration depth.In contrast,a higher DRXed ratio and smaller precipitates along the DRXed grain boundaries facilitate the formation of a more uniform corrosion morphology with a shallow penetration depth.

Fig.14 shows the variation in the maximum penetration depth as a function of the size of theβ-Mg17Al12precipitates formed at the DRXed grain boundaries for the different ECAPed alloys.Fig.14 shows a linear correlation between the maximum penetration depth and the size of theβ-Mg17Al12precipitates formed at the DRXed grain boundaries,in which the size of the precipitates becomes smaller with increasing deformation strain,and these finer precipitates result in lower penetration depths.As explained above,corrosion in the DRXed regions occurs through microgalvanic coupling between theβ-phase along the grain boundaries and the adjacentα-Mg matrix with the preferential dissolution of the matrix.This microgalvanic corrosion process results in undermining and subsequent detachment of theβ-phase from the alloy,thus leaving behind corrosion pits in the attacked regions.As a result,the larger the precipitates,the wider and deeper the corrosion pits.These observations therefore confirm that increasing the level of deformation strain during ECAP processing of AZ91 alloy promotes the formation of shallow corrosion pits upon immersion of the alloy in chloride containing media.

Fig.13.Topography and roughness profiles of the ECAPed alloys with(a)N=1 and(b)N=4 after immersion in 3.5wt% NaCl for 24h.

Fig.14.Maximum penetration depth of the ECAPed alloys after 24h of immersion in 3.5wt%NaCl solution against size of theβ-Mg17Al12 precipitates formed at the grain boundaries of the DRXed grains.

3.2.5.Influence of dynamic recrystallization and refinedprecipitates on the corrosion rate

Fig.15(a)shows the corrosion rates obtained from weight loss measurements of the ECAPed alloys with different numbers of ECAP passes after immersion in 3.5wt% NaCl solution for 7 days.Corrosion rate of the as-cast alloy was also included in Fig.15(a)to evaluate the corrosion resistance performance of the ECAPed alloys against the as-received cast alloy.Fig.15(a)clearly shows the significant improvement in corrosion resistance of the as-received cast alloy after ECAP processing.After the first ECAP pass,the corrosion rate of the AZ91 alloy decreased by 76.5% with respect to the ascast alloy and a total decrease of 91.1% was achieved after ECAP processing for four passes.These findings demonstrate the effectiveness of ECAP in enhancing the corrosion resistance of AZ91 alloys.The inserted corroded morphologies in Fig.15(a)show substantial loss of material in the as-cast alloy that seems to propagate from the periphery of the sample to the sample interior.In contrast,the loss of material is significantly reduced after increasing the number of ECAP passes.Fig.15(a)also shows that the corrosion rate appears to exhibit a linear relationship with the number of ECAP passes,in which the corrosion rate decreases with increasing number of passes.These results further confirm that increasing the level of deformation strain improves the corrosion resistance of the ECAPed alloys.The corrosion rate decreases from 4.43 mmpy after the first ECAP pass to 1.68 mmpy after the fourth pass.The decrease in corrosion rate with increasing strain is mainly attributed to the presence of fineβ-Mg17Al12precipitates along the DRXed grain boundaries,which promote shallow pitting during the micro-galvanic corrosion process and lead to less material loss.The lower corrosion rate with increasing ECAP passes is also associated with an increase in the DRX ratio and volume fraction of well-distributed fineβ-Mg17Al12precipitates,which can prevent the development of severe and accelerated localized corrosion attack.Instead,these microstructural features promote a more uniform corrosion.Furthermore,the corrosion resistance of AZ91 alloys after severe plastic deformation has been reported to be enhanced due to the formation of uniformly distributed fineβ-Mg17Al12precipitates,which induce the formation of an Alrich oxide layer at the alloy surface that is more protective against chloride species than the traditional MgO/Mg(OH)2layer[43].The Al-rich oxide layer has been speculated to form by redeposition ofβ-Mg17Al12precipitates on the alloy surface during the micro-galvanic corrosion process[43].

Vickers hardness of the polished ECAPed samples was plotted against the corrosion rates obtained from weight loss measurements to analyze the influence of deformation strain on both,hardness and corrosion resistance.From Fig.15(b),it is noticed that ECAP processing was effective in simultaneously improving the hardness and corrosion resistance of the as-cast AZ91 alloy.A detailed analysis of the enhanced mechanical properties and corrosion resistance of AZ91 alloys after different fabrication and processing methods including casting,solution heat treatment,aging,ECAP,and postaging after ECAP was reported in our previous study[44].Fig.15(b)also shows that hardness and corrosion resistance are gradually enhanced as increasing the level of strain.The hardness increases from 73.8 HV after the first ECAP pass to 89.1 HV after the fourth pass.The improvement in hardness with increasing deformation strain is mainly attributable to grain boundary strengthening and precipitation hardening[6,7,22,41].The grain boundary strengthening is associated with substantial grain refinement during ECAP processing,which is promoted by DRX and the pinning effect of theβ-Mg17Al12precipitates at the DRXed grain boundaries,which can restrict grain growth.In addition,precipitation hardening is associated with the formation of a high volume fraction of fineβ-Mg17Al12precipitates due to SIDP during ECAP processing.Smaller grain sizes are well known to enhance the grain boundary strengthening according to the Hall–Petch relationship,whereas finer precipitates promote more effective strengthening via the Orowan mechanism[22,29,45].Thus,the combination of smaller DRXed grain size,higher DRX ratio,and a higher volume fraction of uniformly distributed fineβ-Mg17Al12precipitates with increasing ECAP passes improves the hardness(and strength)of the ECAPed alloys.The simultaneous improvement of hardness and corrosion resistance of the ECAPed AZ91 alloys at high levels of severe plastic strain provides a potential opportunity to use magnesium alloys in a wide variety of engineering applications.

Fig.15.(a)Corrosion rate obtained from weight loss measurements of the as-received cast alloy and the ECAPed alloys immersed in 3.5wt% NaCl solution for seven days.(b)Vickers hardness of the polished samples against corrosion rate for the as-cast and ECAPed samples showing the simultaneous improvement in hardness and corrosion resistance as increasing the number of ECAP passes.

3.2.6.Influence of dynamic recrystallization and dynamic precipitation on the passive behavior

Grain refinement has been widely reported to improve the corrosion resistance of magnesium alloys[3,4,46].Most of these studies agree that the improvement in corrosion resistance with decreasing grain size is mainly attributable to the formation of a more coherent,uniform,and protective oxide film,which is more resistant to localized breakdown by aggressive species.To analyze the influence of DRXed grain size on the corrosion behavior of ECAPed AZ91 alloys,we performed electrochemical measurements including OCP,potentiodynamic polarization,and EIS on the ECAPed alloys after different numbers of ECAP passes in a diluted 0.05M NaCl solution.The diluted NaCl solution enables identification of passivation regions as well as differences in the breakdown potential as the number of ECAP passes increases(i.e.,decreasing the DRXed grain size).Fig.16(a)shows the OCP evolution of the ECAPed alloys during 1h of immersion in 0.05M NaCl solution before EIS and potentiodynamic polarization testing.A similar OCP trend is observed for the different ECAPed alloys,in which the OCP values rapidly increase after the first few minutes of immersion(～10min),then stabilize after longer immersion times.The initial increase in OCP has been associated with the formation of an oxide film on the alloy surface,whereas the plateau at longer immersion time suggests that the oxide film is protective and stable,and there is no evidence of localized breakdown under the immersion conditions[47].After 1h of immersion,the OCP values of the ECAPed alloys are more positive as the number of ECAP passes increases;Table 2 shows that the OCP value of the ECAPed alloy after the first pass is?1.540V vs.SCE,whereas the OCP value of the ECAPed alloy after the fourth pass is?1.492V vs.SCE,corresponding to an increase of approximately 50mV.Previous studies have reported that more positive OCP values are associated with more protective oxide films[47,48].In that sense,theresults shown here indicate that increasing the level of strain promotes the formation of a more protective oxide film.

Table 2Electrochemical parameters obtained from OCP and polarization measurements of ECAPed AZ91 alloys after different ECAP passes immersed in 0.05M NaCl solution.

Fig.16.(a)Open circuit potential(OCP),(b)potentiodynamic polarization and EIS measurements showing(c)the Nyquist representation and the(d)the Bode plot of ECAPed alloys with various number of passes immersed in 0.05M NaCl solution.

To further investigate the protective properties of the oxide films formed on the ECAPed alloys,we performed potentiodynamic polarization measurements in diluted 0.05M NaCl solution.Fig.16(b)shows the potentiodynamic polarization curves of the ECAPed alloys immersed in 0.05M NaCl solution.In addition,Table 2 summarizes the values of corrosion potential(Ecorr),corrosion current density(icorr),breakdown potential of the oxide film(Eb),and magnitude of the passive region(Eb–Ecorr)derived from Tafel extrapolation.As shown in Table 2,theEcorrvalues are slightly more positive,and theicorrvalues become smaller with increasing strain level.Theicorrvalues decrease from 1.530μA/cm2for the ECAPed alloy after the first pass to 1.021μA/cm2for the ECAPed alloy after the fourth pass.According to Faraday’s law,lowericorrvalues indicate that the corrosion rate of the ECAPed alloys decreases with increasing strain level,in agreement with the weight loss measurements reported in Fig.15.The cathodic and anodic branches of the polarization curves in Fig.16(b)represent the hydrogen evolution and the dissolution of Mg,respectively.Fig.16(b)clearly shows that the ECAPed alloys exhibit a passive behavior in the anodic branch,thus corroborating the formation of an oxide film on the alloy surface.Furthermore,Fig.16(b)shows that the breakdown potential of the oxide film increases with increasing strain.From Table 2,Ebof the ECAPed alloy after the first pass is?1.308V vs.SCE,whereas theEbof the ECAPed alloy after the fourth pass is?1.042V vs.SCE.According to these results,there in an increase in the magnitude of the passive region(Eb–Ecorr)with increasing strain level,where the passive region of the ECAPed alloy after the first pass has a magnitude of 144mV,whereas this magnitude increases to 392mV after the alloy is extruded for four passes.The inset in Fig.16(b)also shows that the anodic current density decreases with increasing strain level.These results further confirm that the protective properties of the oxide film are enhanced with increasing strain level.Consequently,the substantial grain refinement of the DRXed grains and the higher volume fraction of DRXed grains observed with increasing strain level may promote the formation of a more protective oxide film with higher resistance to localized breakdown by aggressive species.

A similar cathodic response was observed for the ECAPed alloys,thus indicating that the cathodic reaction rate(i.e.,hydrogen evolution rate)is not significantly influenced by the level of strain.As shown in the inset in Fig.16(b),these is a slight increase in the cathodic current density with increasing strain level;this behavior has been associated with the increase in the volume fraction ofβ-Mg17Al12precipitates with increasing strain level,thus providing sites for cathodic reactions and resulting in faster cathodic reaction kinetics[4].However,Table 2 shows an overall decrease in theicorrvalues with increasing strain level;these results suggest that the decrease in the anodic current density with increasing strain level predominates over the slight increase in cathodic current density.Thus,the corrosion process of the ECAPed alloys in the diluted NaCl solution is controlled by the anodic reaction kinetics(i.e.,protective properties of the oxide film).

Fig.17.Variation of Eb–Ecorr of the ECAPed alloys with different number of ECAP passes against(a)reciprocal square root of grain size and(b)volume fraction ofβ-Mg17Al12 precipitates at the DRXed grain boundaries.

Ralston et al.[49]have demonstrated a linear correlation between corrosion current density and the reciprocal square root of grain size(d?1/2),analogously to the Hall–Petch relation;icorr=A+Bd?1/2,in which the constantAis associated with the nature of the environment,and the constantBis associated with material properties.According to this relationship,and for negative values ofB,icorrdecreases with decreasing grain size.The authors have attributed the enhanced corrosion resistance with decreasing grain size to the formation of a more protective oxide film.In this study,a similar correlation was found;however,instead of correlatingicorrwith the grain size,a relationship between the magnitude of the passive region(Eb–Ecorr)and grain size was analyzed.The magnitude of the passive region(Eb–Ecorr)obtained from the anodic branch of the polarization curve is thought to be a better parameter for analyzing the protective properties of the oxide film as a function of grain size.Fig.17(a)shows the variation inEb–Ecorras a function of grain size.An excellent linear correlation is observed betweenEb–Ecorrandd?1/2,in which decreasing the grain size(i.e.,increasing the number of ECAP passes)results in a higher passivation region and therefore in a more protective oxide film with a higher resistance to localized breakdown by aggressive species.The high volume fraction of grain boundaries(i.e.,smaller grain size)in ultra-fine grained magnesium alloys provides nucleation sites for oxide film formation,thus resulting in a more compact and adherent protective layer[48,50,51].Other authors have proposed that ultra-fine grained microstructures can relieve stresses induced at the substrate/oxide interface,owing to the volume mismatching between the Mg substrate and the MgO layer,thereby resulting in a MgO layer that is less susceptible to cracking and consequently more resistant to migration of aggressive species[3,4].Furthermore,for Mg–Al alloys with high aluminum content,such as the AZ91 alloy,grain size is not the only factor that can influence the properties of the oxide film[4].The high volume fraction ofβ-Mg17Al12precipitates at the DRXed grain boundaries is also believed to play an important role in enhancing the protective ability of the oxide film,through promoting the formation of an aluminumenriched oxide(Mg,Al)xOyfilm with better protective properties than those of the semi-protective MgO/Mg(OH)2oxide film and with higher stability over a wide range of pH values[43,52].Fig.17(b)shows the variation inEb–Ecorras a function of the volume fraction ofβ-Mg17Al12precipitates at the DRXed grain boundaries for the different ECAPed alloys.The magnitude of the passive region increases with increasing the volume fraction of precipitates(i.e.,increasing strain level).Although the relationships among these parameters appear to be more complex than that betweenEb–Ecorrandd?1/2,increasing the volume fraction of precipitates clearly contributes to the formation of a more protective and stable passive film with a higher resistance to localized breakdown.Therefore,the substantial grain refinement coupled with a high volume fraction of well-distributedβ-Mg17Al12precipitates is likely to have a synergistic effect in promoting the formation of a more protective oxide film,and this synergistic effect can overcome the negative effect of theβ-Mg17Al12precipitates in accelerating the cathodic reaction kinetics.

Fig.16(c)and(d)shows Nyquist and Bode representations,respectively,of the ECAPed alloys immersed in 0.05M NaCl solution.EIS measurements were performed to support the potentiodynamic polarization results regarding the improved protective ability of the oxide film with increasing strain level.As shown in Fig.16(c),the EIS spectra of the ECAPed alloys exhibit two capacitive loops:one capacitive loop from high to intermediate frequencies(60 kHz–160 mHz)and a capacitive loop at low frequencies(160 mHz–10 mHz).Furthermore,the radii of these capacitive loops increase with increasing strain level.According to previous studies,the high-frequency capacitive loop has been associated with the protective properties of the oxide film,whereas the low-frequency capacitive loop has been attributed to charge transfer during electrochemical reactions at the metal/electrolyte interface[52,53].The larger the radius of the high-frequency loop,the greater the protective ability of the oxide film.Similarly,the greater the radius of the lowfrequency loop,the higher the resistance of the Mg alloy to charge transfer processes.Therefore,Fig.16(c)shows that increasing the level of strain results in the formation of a more protective oxide film.The enhanced oxide film formation with increasing strain level subsequently improves the corrosion resistance of the Mg alloy by blocking active anodic and cathodic sites,such that charge transfer processes are delayed at the alloy surface.The Bode plots in Fig.16(d),showing the impedance magnitude(|Z|)and the phase angle against frequency,further support these observations.Typically,the impedance magnitude at the lowest frequency(|Z|0.01Hz)is associated with the total impedance of the system,and the greater the|Z|0.01Hzvalue,the higher the corrosion resistance of a material.In this sense,Fig.16(d)shows that the|Z|0.01Hzvalues of the ECAPed alloys increase with increasing strain level,thus indicating that ECAP processing at higher strain levels improves the corrosion resistance of the ECAPed alloys.The phase angle curve confirms the presence of the two time constants observed from the Nyquist representation.Furthermore,the phase angle of the high-frequency time constant decreases with increasing strain level.These results also indicate that the protective properties of the oxide film are enhanced with increasing strain level.

Table 3Electrical parameters obtained from fitting the EIS data of the ECAPed alloys immersed in 0.05M NaCl solution.

The EIS spectra of the different ECAPed alloys were fitted with the equivalent electrical circuit shown in Fig.16(c)to further explore the electrochemical response of these ECAPed alloys.In this circuit,Rscorresponds to the resistance of the electrolyte,CPEoxandRoxdescribe the capacitance and the resistance of the oxide film,respectively,and CPEdlandRctrepresent the double layer capacitance and the charge transfer resistance,respectively,of the charge transfer process at the Mg alloy/electrolyte interface.Constant phase elements(CPEs)were used instead of capacitance values to account for nonideal capacitive behavior due to heterogeneities in the system(e.g.,surface roughness,electrode porosity,non-uniform potential,and current distribution)[54].The CPE element is defined in terms of the parametersTandn;Tcorresponds to the CPE magnitude,and n is an empirical exponent that varies between 0 and 1.Forn=0,the CPE element becomes a pure resistor,and forn=1,the CPE element is identical to a pure capacitor.As seen in Fig.16(c)and(d),there is good agreement between the EIS experimental data and the fitted results obtained from the equivalent circuit.The different electrical parameters obtained after the EIS fitting process are reported in Table 3.From these results,the resistance of the oxide film(Rox)clearly increases with increasing strain level,from 12.36 k?·cm2for the ECAPed alloy after the first pass to 17.56 k?·cm2for the ECAPed alloy after the fourth pass,thus corroborating that increasing the strain level during ECAP processing results in ECAPed alloys that form more protective oxide films with a higher resistance to aggressive species.Accordingly,Table 3 shows that the capacitance of the oxide film(CPE-Tox)decreases with increasing strain level.This parameter has been associated with the thickness of the oxide film,in which lower values of CPE-Toxare associated with relatively thicker and more compact films[8,47,48].Therefore,the lower CPE-Toxvalues with increasing strain level indicate the formation of a more compact oxide film that hinders the migration of aggressive species toward the Mg surface.Improvement in the protective properties of the oxide film is expected to subsequently improve the corrosion resistance of the Mg alloy by preventing aggressive species from reaching the metallic surface.This is indeed the case in this study;Table 3 indicates an increase in the charge transfer resistance(Rct)and a decrease in the double layer capacitance(CPE-Tdl)with increasing strain level.The increase in theRctvalues with increasing strain level indicates that the ECAPed alloys with severe plastic strain exhibit higher resistance to electrochemical reactions induced by a corrosive environment.Furthermore,the decrease in the CPE-Tdlvalues with increasing strain level suggests that the active area available for charge transfer processes decreases with increasing strain level.

These EIS results and equivalent circuit analysis correspond well with the potentiodynamic polarization results confirming that increasing the number of ECAP passes leads to formation of more protective oxide films that are more effective in blocking active sites for the development of electrochemical processes at the alloy surface.Consequently,the enhanced protective ability of these oxide films can further improve the corrosion resistance of the ECAPed alloys.As mentioned above,the superior protective ability of the oxide film on the ECAPed alloys with higher strain levels can be associated with a potential synergistic effect between the significant grain refinement coupled and a high volume fraction of well-distributedβ-Mg17Al12precipitates.

4.Conclusions

The corrosion susceptibility of recrystallized and unrecrystallized regions and the combined effect of dynamic recrystallization and dynamic precipitation on the electrochemical behavior of partially recrystallized ECAPed AZ91 alloys at different strain levels was investigated.The main conclusions from this study are summarized as follows:

?Immersion testing of the ECAPed alloys during short and long immersion times in 3.5wt% NaCl solution showed that corrosion is primarily initiated and propagated in the DRXed grains through microgalvanic coupling between theβ-Mg17Al12precipitates formed at the DRXed grain boundaries and the adjacentα-Mg matrix.These results were confirmed by SKPFM and local PDP measurements in which the DRXed region was found to exhibit a higher Volta potential difference and higher corrosion current density than the un-DRXed region.The higher corrosion susceptibility of the DRXed region is mainly attributed to the greater microchemistry difference between the precipitates and the matrix and the weaker basal texture of DRXed grains.

?The corrosion morphology of the ECAPed alloys changes with increasing deformation strain from highly localized and deep at low strain levels to more uniform and shallower at higher strain levels.The increase in strain level also leads to lower corrosion rates of the ECAPed alloys.The more uniform corrosion with shallow penetration depth at high strain levels is attributed to the higher DRXed ratio and the high volume fraction of well-dispersed fineβ-Mg17Al12precipitates formed at the DRXed grain boundaries.A linear correlation was found between maximum penetration depth and size ofβ-Mg17Al12precipitates.

?Increasing the level of deformation strain promotes the formation of a more protective oxide film on the ECAPed alloys with higher resistance to localized breakdown by aggressive species.The greater protective properties of the oxide film with increasing strain level is associated with substantial grain refinement that promotes the formation of a more compact and adherent surface protective layer.Furthermore,the high volume fraction of well-dispersedβ-Mg17Al12precipitates at the DRXed grain boundaries can also induce the formation of an aluminum-enriched oxide layer with more protective ability than the MgO/Mg(OH)2layer.Correlations betweenEb–Ecorrandd?1/2andEb–Ecorrand volume fraction of precipitates were established.

?A simultaneous improvement in corrosion resistance and hardness was achieved as increasing the level of deformation strain,these results serve as a platform for future alloy design.

Data availability

The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study.

Funding

This research did not receive any specific grant from funding agencies in the public,commercial,or not-for-profit sectors.

Declaration of Competing Interest

There is no conflict of interest.

Acknowledgments

The authors acknowledge Robert Barber for assisting during the ECAP processing of the AZ91 alloys and Michael Elverud for helping in the machining and heat treatment of the AZ91 samples.The authors would also like to thank Abhinav Srivastava and Dr.Ahmad Ivan Karayan for their technical assistance and their scientific contribution in analyzing the data.