Yuzhou Du ,Xin Wang ,Dongjie Liu ,Wanting Sun ,Bailing Jiang

a School of Materials Science and Engineering,Xi’an University of Technology,Xi’an 710048,China

b Shaanxi Province Engineering Research Center for Magnesium Alloys,Xi’an University of Technology,Xi’an 710048,China

cSchool of Materials Science and Engineering,Dongguan University of Technology,Dongguan 523808,China

Abstract The present study investigates the corrosion behavior of a Mg-Zn-Ca-La alloy.Results indicate that the corrosion resistance is enhanced by rolling deformation and subsequent annealing treatment significantly .The microstructure,distribution of second phase and grain orientation are the main factors affecting corrosion properties.The second phase with semi-continuous network in the as-cast alloy gives rise to the penetration of corrosion into the matrix.However,the dispersed second phase after deformation results in uniform corrosion.The formation of dense corrosion layer at the initial stage of corrosion is favorable for the improvement of corrosion resistance.The sample with 〈0001〉plane oriented towards surface exhibits good corrosion resistance,which is ascribed to the homogeneous microstructure and strong basal texture.

Keywords: Mg-Zn-Ca-La;Rolling;Second phase;Texture;Corrosion properties.

1.Introduction

Nowadays,the development of low-emission vehicles transportation is urgent to reduce the usage of fossil fuel and CO2emissions,where weight reduction is critical.Magnesium and its alloys with high specific strength,exceptional machinability,excellent castability [1],are considered to be a promising engineering metal to reduce the weight of traffi vehicles.However,commercial Mg alloys still suffer from many challenges,such as low elastic modulus[2],poor formability [3],relatively low strength [4] and inferior corrosion resistance [5].Thus,the modern advanced technologies have been employed to overcome the existed shortcoming of Mg alloys [6].For instance,a Mg-3Al-1Zn-0.3Mn (wt%) alloy with the yield strength of 380MPa in both tension and compression was obtained by extrusion at 175 °C [7];the texture of a Mg-1Zn-1Gd (wt%) alloy was modified through differential speed rolling and electro-pulse treatment [8].Although obvious progresses in properties improvement of Mg alloys have been made,Mg alloys still have inferior corrosion resistance by contrast to their competitors such as Al alloys[9].In this regard,developing Mg alloys with high corrosion resistance attracts researchers’ attentions in the past decades[10,11].

Considering the control of Mg alloy corrosion,surface treatment[12]or adding appropriate alloying elements[13,14]is proven to be effective methods.The introduction of protective coatings by surface treatment is accompanied with cost increase.Thus,enhancing corrosion resistance of Mg alloys via alloying remains preferentially adopted by researchers[15].In the past decades,many kinds of alloying elements were added in Mg to improve the mechanical properties and corrosion resistance [16,17].Mg-Zn binary alloy system exhibited better corrosion resistance compared with pure Mg when Zn content is below 4wt%,which was related to the formation of Zn oxides as an inner layer where the Mg (OH)2act as top layer during immersion tests [18].However,more Zn content had a detrimental effect on corrosion property due to the effects of second phase [19].Mg-Ca alloy was considered to be a biodegradable material due to its superior biocompatibility to human body [20],and Ca concentration in Mg alloys should be controlled to be less than 1wt% [21].Currently,the corrosion behavior of Mg-Zn-Ca alloys was widely investigated in previous literatures [22-24].Small addition of La resulted in the formation of La oxide,which was beneficial for the improvement of corrosion resistance [25].Therefore,the present study fabricated a Mg-3Zn-0.3Ca-0.La(wt%),which might be a promising corrosion resistant Mg alloy.Our previous investigation has suggested that Mg-3Zn-0.3Ca-0.4La (wt%) alloy was a low-cost high-strength Mg alloy in [26].However,the corrosion behavior of this kind Mg alloy is still unclear.

Thermo-mechanical process is capable of refining microstructure[27],modifying distribution of second phases and thus affecting corrosion properties of Mg alloys [28].For example,the corrosion resistance of Mg-14Li-3Al (wt%) alloy was enhanced through refining the microstructure by severe plastic deformation [29];indirect extrusion reduced the nobility of cathodic Al5Fe2particles of Mg-8Sn-1Zn-1Al (wt%)alloy,which retarded microgalvanic corrosion betweenα-Mg and Al5Fe2and enhanced corrosion resistance [30];a Mg-Ca alloy with uniform dispersion of Mg2Ca phase was obtained through high-ratio differential speed rolling,which reduced the occurrence of microgalvanic corrosion and gave rise to a significant enhancement of corrosion resistance[31].For Mg-Zn-Ca-La alloy,previous investigations mainly focused on microstructural evolution and mechanical properties improvement [32,33].The corrosion behavior of Mg-Zn-Ca-La alloy has seldom been investigated.Additionally,high deformation amounts promoted dynamic recrystallization and increased the microstructural homogeneity [34],and further improved corrosion resistance of Mg alloys [35].Therefore,a Mg-Zn-Ca-La alloy was subjected to rolling deformation with a high deformation amounts and annealing treatment,and then the microstructure and corrosion behavior were investigated in the present study.

2.Materials and methods

2.1.Materials

High-purity Mg,Zn and Mg-15% Ca,Mg-30% La master alloys were selected as the raw materials to fabricate Mg-Zn-Ca-La alloy under the protection of a mixed gas atmosphere of SF6 and CO2 by semi-continuous direct-chill casting process.The detailed procedure for casting could be found in our previous investigation [26].The chemical composition was determined to be Mg-3Zn-0.3Ca-0.4La (wt%)by X-ray fluo rescence spectrometry(XRF).The as-fabricated casting ingots were homogenized at 400°C for 12h.Then the samples were machined into slabs with dimensions of 50mm×50mm with a thickness of 8mm for rolling experiments.Before rolling deformation,the samples were heated to 350 °C and held for 15min.Rolling deformation was carried out at 350 °C with a 6% reduction per pass for total 6 passes,which gave rise to a total reduction of approximately 70%.After each rolling pass,the rolling samples were reheated to 350 °C and held for 10min.The rotation speed of the rollers with a diameter of 175mm was maintained at 32rpm during the rolling process.After rolling deformation,the samples were annealed at 300 °C for 0.5h to reduce the internal stress and obtain fully homogeneous microstructure.

2.2.Microstructure characterization

In the present study,the ND plane of the sample after rolling deformation and annealing treatment was referred to the plane perpendicular to normal direction (ND),and TD plane was referred to the plane perpendicular to transverse direction (TD).The surface of the samples were firstly ground by 3000 grit SiC emery paper and cleaned with ethanol,and then mechanically polished with diamond paste.The polished samples were finally ultrasonically cleaned in ethanol and dried with a jet of compressed air.After that,the samples were etched by acetic picral (5ml acetic acid+6g picric acid+10ml H2O+100ml ethanol).The microstructure was observed by Olympus DP11 optical microscopy (OM),and Quanta 200FEG scanning electron microscopy (SEM)equipped with energy-dispersive spectrometer (EDS) under the voltage of 30kV was used for microstructure and constitutional analysis.The grain sizes were measured using the software of Image-Pro-Plus 5.0 with the grain number more than 500.The incomplete (0002),(10-10),(10-11) and (10-12) pole figure of the ND and TD samples were measured on an X’PERT PRO MPD X-ray diffractometer,and the obtained incomplete poles figure were used to calculate full pole figure with the software of MTEX 5.1.1 toolbox [36].

2.3.Immersion tests

The as-cast and ND samples for immersion tests were machined into 10mm×10mm×2mm,and the TD sample was machined into 50mm×2mm×2mm.Afterwards,the samples were mounted into epoxy resin with the exposed surface area of 1 cm2.Prior to the immersion testing,the weight of the samples were accurately measured.The samples were immersed in 3.5wt% NaCl solution at ambient temperature.The test electrolyte herein was 3.5% NaCl solution prepared by deionized water.An inverted burette and funnel set up were utilized to collect the hydrogen gas during the immersion,shown in Fig.1.The hydrogen volume was recorded after immersion for different times to evaluate the corrosion properties.The corrosion products on the surface of the samples immersion for different times were removed using a solution composing of 200g/L CrO3,20g/L Ba(NO3)2and 10g/L AgNO3according to the standard of GBT 16545-2015.Then the corroded samples were ultrasonically cleaned in ethanol and dried with warm airflow.The weight loss after immersion for 80h were measured by a FA1004B analytic balance,and the corrosion morphology was characterized by SEM.

Fig.1.The schematic diagram of hydrogen evolution measurements.

2.4.Electrochemical tests

The specimens for electrochemical tests were sealed with epoxy resin with an exposed area of 1 cm2.A PARSTAT400 electrochemistry workstation was used for electrochemical measurements at room temperature.A three-electrode cell with a platinum counter electrode,a saturated calomel reference electrode (SCE) and the specimen as working electrode were used in the present study.The test electrolyte herein was 3.5% NaCl solution prepared by deionized water at ambient temperature.The specimen was firstly immersed in 3.5% NaCl solution for 30min to reach a stable open circuit potential.Then the potentiodynamic polarization was measured with the sweep rate of 1mV/s.The obtained polarization curves were used to calculate corrosion potential (Ecorr)and corrosion current density (icorr) by Tafel fitting The corrosion rate,Pi(mm/y),was calculated using the following equation:

whereKis a constant(8.96×10-3,n is the number of valence electrons (2 for Mg),andicorris the corrosion current density(μA/cm2),M is relative atomic mass,ρis the density of Mg.

The electrochemical impedance spectroscopy (EIS) was performed with the scan frequency ranging from 105Hz to 0.01Hz and amplitude of 10mV with respect to the OCP.The electrochemical tests were repeated at least three times to ensure the reproductivity.The software of ZSimpwin was used and EIS spectra for the as-cast and ND samples were analyzed using the equivalent circuit presented in Fig.2a.Spectra for TD sample were analyzed using the equivalent circuit of Fig.2b.In these circuits,Rsis the solution resistance,Rctis the charge transfer resistance,Rfis the resistance of corrosion products,LandRLare inductive and its resistance,CPEfandCPEdlrepresent the capacity of corrosion products and the double layer respectively.The constant phase element CPE replacing pure capacitor is because of the non-uniform distribution of sample surface [37,38].The value ofCPEis determined byYandn,in whichnis the dispersion coefficient forCPE.

3.Results

3.1.Microstructure

Fig.3 shows the backscattered electron (BSE) images and EDS results of the as-cast Mg-Zn-Ca-La alloy.The second phase along grain boundaries formed semi-continuous network (Fig.3a) in the as-cast alloy.It should be noted that there existed at least two types of second phases according to the brightness contrast:phase 1 and phase 2 shown in Fig.3b.Also,some spherical second phases distributed at grain interior were detected (Fig.3c).The EDS results indicated that the phase at triple junction (phase 1) and the phase at grain interior (phase 3) contained Mg,Zn,Ca and La (Fig.3d and f).However,the phase at grain boundaries (phase 2) contained 84.4 at.% Mg,11.36 at.% Zn,4.00 at.% Ca and only small amount of La(0.22 at.%)(Fig.3e).It suggested that the as-cast Mg-Zn-Ca-La alloy contained two types of second phases,which was previously confirmed to be Ca2Mg6Zn3and Thase consisting of Mg,Zn,La and small amounts of Ca [26].The volume fraction of second phase for the as-cast alloy was estimated to be 3.9%.

Fig.4 shows the optical and SEM images of the TD and ND samples.A homogeneous microstructure was observed for the TD and ND samples (Fig.4a and b),and the average grain size was approximate 8 μm for the two samples,indicating that recrystallization occurred after rolling deformation and subsequent annealing treatment.From Fig.4c and d,it could be seen that the undissolved second phase was fragmented and elongated along rolling direction.The volume fraction of second phases in TD and ND sample was about 3.5% through calculating the area fraction of second phases,which was less than the as-cast counterpart.It indicated that part of the second phase was dissolved into the matrix during homogenization treatment.The distribution of second phases in the TD and ND samples differed.The second phases were elongated and distributed align rolling direction for TD sample (Fig.4c),while the second phase in ND sample seemed that the initial second phase with semi-continuous network was narrowed and elongated along rolling direction (Fig.4d).Fig.5 shows the EDS results of the second phase in the annealed alloy.It could be seen that the second phase of TD sample was mainly composed of Mg,Zn and La.Only small amounts of Ca was detected in the second phase (Fig.5b),suggesting that Ca2Mg6Zn3was dissolved into the matrix during homogenization and the element of Ca was mainly existed in the form of solute atoms.

Fig.2.Equivalent circuit models of the impedance spectra for (a) the as-cast and ND sample and (b) TD sample.

Fig.3.The BSE images of the as-cast alloy:(a) low magnification (b) and (c) the enlargement of the region A and B in a,and the EDS results of the point(d) 1,(e) 2 and (f) 3.

Fig.6 presents the pole figure of the ND and TD samples.A typical basal texture was observed for the ND sample (Fig.6a).The basal texture split into two poles along rolling direction,which was a typical texture type in asrolled Mg-Zn-RE alloy [39] and Mg-Gd alloy [40].The formation of this type texture was related to the basal slip and tension twinning in Mg alloys [41].The basal texture implied that most of the basal planes were paralleled to rolling plane,i.e.,the ND sample contained most basal planes oriented towards surface.However,For the TD sample,it could be seen that (10-10) plane was perpendicular the TD(Fig.6b),indicating that a prismatic texture formed in TD sample.The texture intensity for the ND and TD samples was similar.

Fig.4.Optical images of (a) TD and (b) ND samples showing homogeneously DRXed microstructure,and SEM images of (c) TD and (d) ND samples showing obvious difference of second phase distribution.

Fig.5.(a) SEM morphologies and (b) EDS results of the second phase in a of the TD sample.

3.2.Corrosion properties

3.2.1.Immersion tests

Fig.7a shows the hydrogen evolution of the Mg-Zn-Ca-La alloy immersed in 3.5wt% NaCl solution at room temperature.The ND sample exhibited a much lower hydrogen evolution compared with the as-cast alloy.However,the TD samples exhibited much higher hydrogen evolution.Close observation of the as-cast alloy and ND samples could be found that the two alloys exhibited similar hydrogen evolution rate at the initial stage of immersion test (before 24h immersion).However,the hydrogen evolution rate of the as-cast alloy was increased obviously with the increase of immersion time,while the ND sample exhibited a slower hydrogen evolution rate.This indicated that rolling deformation and subsequent annealing treatment enhanced the corrosion resistance of Mg-Zn-Ca-La alloy.The weight loss of the samples immersed for 80h was presented in Fig.7b.The weight loss of ND sample was smaller than that of the as-cast sample,but the TD sample exhibited a much higher weight loss.This was consistent with the hydrogen evolution.

Fig.6.The pole figure of the (a) ND and (b) TD samples.

Fig.7.(a) Hydrogen evolution with immersion time and (b) weight loss of the samples immersed in 3.5wt% NaCl for 80h.

Fig.8 shows the corrosion morphology of the Mg-Zn-Ca-La alloys immersed in 3.5%NaCl solution for different times.After immersion for 1h,the TD sample exhibited obvious corrosion morphology,while only small amounts of pitting corrosion were observed for the as-cast and ND samples.With the increase of immersion time,the TD sample was corroded much more severe than the as-cast and ND samples,which was consistent with hydrogen evolution.However,it seemed that the ND sample corroded faster than the as-cast alloy by contrasting the morphologies of the two alloys for samples immersion for 6h.For samples immersion for 24h,which was the point of the as-cast began producing hydrogen at a higher rate than ND sample (Fig.7a),the corrosion morphology of the as-cast and ND samples differed little (Fig.8a and b).The surface of the two samples were almost corroded.In order to reveal the corrosion behavior of the as-cast and ND sample,the cross section morphology after immersion in 3.5wt% NaCl for 3h was shown in Fig.9.It could be clearly seen that the corrosion depth of the as-cast alloy was higher than that of the ND sample.The corrosion depths for the as-cast,ND and TD samples were about 30 μm,9 μm and 64 μm,respectively.Additionally,a higher surface roughness was observed for the as-cast alloy by contrast to the ND sample.It also should be noted that the ND sample was corroded much more uniform than the as-cast alloy (Fig.9a and b).

Fig.8.The corrosion morphology of the (a) as-cast,(b) ND and (c) TD samples after immersion in 3.5 wt% NaCl for different times,showing the corrosion evolution progress.

Fig.9.The cross-section morphology of the (a) as-cast,(b) ND and (c) TD samples after immersion in 3.5 wt% NaCl for 3h,indicating different corrosion depths for the three samples.

3.2.Electrochemical measurements

3.2.1.Polarization tests

Fig.10 shows the potentiodynamic polarization test results of the Mg-Zn-Ca-La alloys.The corrosion potential (Ecorr)and corrosion current density (icorr) values were determined from the potentiodynmic polarization curves and summarized in Table.1.Compared with the as-cast sample,there existed a positive shift in theEcorrfor the ND sample,which indicated that rolling deformation enhanced the stability of Mg-Zn-Ca-La alloy.The corrosion ratePicalculated by Eq.(1) was also shown in Table 1.It could be seen that the TD sample exhibited much higher corrosion rate than the as-cast and ND samples,and the corrosion rate of ND sample was smaller than the as-cast sample.This outcome was consistent with theresults of hydrogen evolution rate (Fig.7a).For Mg alloys in aqueous solution,the cathodic polarization curves represented hydrogen evolution,and anodic polarization curves signifie dissolution of Mg [31].The ND sample exhibited a decrease in anodic reaction,which gave rise to the decrease oficorrvalues from 23.4 μA/cm2for the as-cast sample to 16.5 μA/cm2for ND sample.It indicated that rolling deformation and subsequent annealing treatment improved the corrosion resistance of the Mg-Zn-Ca-La alloy.The corrosion current density was much lower than pure Mg (370.7 μA/cm2) [18].Additionally,the corrosion potential was positively shifted compared with pure Mg (-1.78V)[18,42],indicating that Mg-Zn-Ca-La alloy was more stable thermodynamically by contrast to pure Mg.One may noticed that the as-cast sample exhibited a little slower cathodic reaction rate than ND sample.The cathodic reaction rate corresponded the hydrogen evolution of the initial stage of immersion test because the polarization tests of the present study were finishe in one hour,during which the as-cast and ND samples showed similar hydrogen evolution(Fig.7).To differentiate the effect of crystal orientation on corrosion resistance,the polarization curve of the TD sample was plotted in Fig.10.The obvious increase of the anodic and cathodic kinetics of the TD sample compared with that of the ND sample resulted in a high corrosion current density(958.6 μA).This results confirmed that crystal orientation was an important factor for corrosion properties of Mg alloys.The as-cast and ND samples showed an inflection point (arrowed in Fig.9) on the anodic polarization curve,which represent the breakdown of the film and always observed in Mg alloys[43].

Fig.10.The potentiodynamic polarization curves for the Mg-Zn-Ca-La alloy immersed in 3.5wt% NaCl solution.

Table 1Fitting results of potentiodynamic polarization curves of Mg-Zn-Ca-La alloys in 3.5wt% NaCl solution.

3.2.2.Electrochemical impedance spectroscopy (EIS)

Fig.11 shows the impedance spectra of the alloys measured in 3.5% NaCl solution.Two types of curves were observed for the three alloys,indicating that different corrosion mechanisms occurred.The as-cast and ND samples shows two capacitive semicircles (Fig.11a),in which the high-frequency domain reflected charge transfer resistance and the mediumfrequency one corresponded to mass transport resistance in the solid phase [44].The ND sample had a larger diameter of high-frequency capacitive loop than the as-cast sample,indicating that a higher charge transfer resistance for the ND sample.However,the TD sample contained a rather small diameter of high-frequency capacitive loop,signifying a poor corrosion resistance,which was consistent with the hydrogen evolution results.However,the TD sample exhibited a very different curve characteristics in the Nyquist plot.In the low-frequency domain,the as-cast and ND samples exhibited diffusion capacitive loop,while the TD sample showed one inductive loop.The existence of inductive loop was reported to be related to pitting corrosion [45].Therefore,it could be inferred that the matrix of TD sample would have severe pitting corrosion at the initial stage of corrosion.The bode-frequency plots (Fig.11b) showed that the ND sample exhibited a higher impedance,demonstrating that rolling deformation improved the corrosion resistance.Additionally,the TD sample exhibited the lowest impedance,which indicated that a poor surface protective effect for the TD sample and the grain orientation affected the corrosion resistance of Mg alloys significantly .

The equivalent circuits presented in Fig.2 were used to fi EIS data.Fig.2a was used to fi data of the as-cast and ND samples,and Fig.2b was used to fi data of the TD sample.The fitte data are shown in Table 2 based on the above equivalent circuit model.The ND sample had a largerRctvalue compared with the as-cast alloy,indicating that rolling deformation improved the corrosion resistance.However,Rctfor TD sample is rather small,suggesting that the grains with prismatic planes are easily to be corroded.Additionally,the valueRfof the ND sample was much larger than that of the as-cast alloy,indicating that the corrosion products on the surface of ND sample was more stable.

Table 2Fitting results of EIS Measurement of Mg-Zn-Ca-La alloys in 3.5wt% NaCl solution.

Fig.11.Impedance plots of Mg-Zn-Ca-La alloys immersed in 3.5 wt% NaCl solution:(a) Nyquist diagram,and (b) Bode plots.

3.3.Corrosion morphology

Fig.12 presents the corrosion morphology of the as-cast and ND samples after immersion in 3.5wt% NaCl for 0.5h.Small amounts of corrosion sites were observed for the ascast and ND samples,and the corrosion seemed to occurred in the vicinity of second phase (Fig.12a1and b1).It indicated that galvanic corrosion existed for the as-cast and ND samples.The magnified morphology suggested that the ascast alloy exhibited corrosion attack around the second phase(Fig.12a2),while the ND sample showed a pancake shaped corrosion morphology (Fig.12b2).It was mainly because the coarse second phase in the as-cast alloy resulted in a large incompatible during corrosion of the matrix.Corrosion products like thin strips on the surface of the as-cast and ND samples were observed from the magnified SEM images (Fig.12a3and b3).Interestingly,the corrosion products of the as-cast alloy showed a relatively loose and porous structure,while a compact and dense corrosion products were detected for the ND sample (Fig.12b3).Therefore,it could be inferred that corrosion products of the ND sample provided more effective protection compared with that of the as-cast alloy,which was consistent with the value ofRfmeasured according to the EIS (Table 2).The ND sample exhibited a much more homogeneous and fine microstructure compared with the as-cast alloy (Fig.2 and 3),which might be the reason for the dense and corrosion products at the initial stage of corrosion for the ND sample [46].EIS measurements demonstrated that the inductive behavior,which commonly existed in pure Mg and AZ31 alloy [47],was not observed for the as-cast and ND samples.It suggested that the film formed on the as-cast and ND sample was different from the film on the conventional Mg alloy.Small addition of Ca in the Mg-Zn-Ca-La alloy would resulted in the formation of Ca(OH)2during immersion tests [48] or CaCO3[22] and La gave rise to the formation of La oxide [25],which was beneficial for obtaining a dense corrosion layer.Consequently,the inductive behavior was not detected for the as-cast and ND samples.

Fig.13 shows the corrosion morphology of the Mg-Zn-Ca-La alloy after immersion in 3.5wt%NaCl for 3h.It could be seen that the as-cast alloy was severely corroded and many corrosion pits were created on the corroded surface(Fig.13a).However,the corrosion of the ND sample was more uniform than that of the as-cast alloy (Fig.13b).The area of the ND sample corroded seemed to be larger than that of the as-cast alloy.However,the corrosion depth of the as-cast alloy was much higher than that of the ND sample.It indicated that the corrosion going into the matrix for the as-cast alloy was more easily,which was consistent with the observation in Fig.9.It is worth noting that the TD sample was severely corroded(Fig.13c),which should be related to the distribution of second phase and grain orientation.

4.Discussion

The microstructure of the as-cast alloy differed from that of the sample after rolling deformation and annealing treatment.The ND sample exhibited a more finel homogeneous microstructure,and the second phase dispersed much more uniform.Additionally,strong basal texture was formed in the ND sample.These microstructural changes should be responsible for the difference of corrosion behavior.

Mg alloys were very sensitive to galvanic corrosion[49,50],which could be induced by other metals adjacent to Mg or second phases in Mg matrix.The role of second phase in Mg alloy during galvanic corrosion depended on the electro-potential relative to Mg matrix.The as-cast Mg-Zn-Ca-La alloy mainly contained Ca2Mg6Zn3or MgZnLa-Ca phase,while only MgZnLa-Ca phase was detected in the ND or TD sample.The previous investigation proved that a T phase with (Mg7Zn3RE) had a relative high Volta potential compared withα-Mg matrix in a Mg-Zn-RE alloy[51].Therefore,the second phase (Ca2Mg6Zn3or MgZnLa-Ca phase) in the Mg-Zn-Ca-La alloy acted as an active cathode,which would form galvanic couple during corrosion process and therefore prompted the dissolution of Mg matrix.Fig.14 shows a typical corrosion surface morphology of the as-cast sample immersion in 3.5wt% NaCl for 6h.It could be clearly seen that corrosion products formed in the vicinity of second phase,indicating the second phase acting as microcathode during corrosion.The size and distribution of second phase had important effects on the corrosion performance of Mg alloys.For the as-cast alloy,the network second phase provided a barrier for corrosion extending forward along the previously corroded paths,which resulted in corrosion preferentially expanding into the grain interior (Fig.13a).For the ND sample,the second phase exhibited much fine distribution,which gave rise to the formation of more micro-galvanic sites.Consequently,a more uniform corrosion morphology was observed (Fig.13b).However,the second phases along initial grain boundaries in TD samples could provide barrier for corrosion extending forward along the surface like the as-cast alloy and was favorable for the corrosion into the interior of Mg matrix.As a result,a high corrosion depth was observed for the TD sample (Fig.8c).

Fig.13.The surface morphology of the (a) as-cast and (b) ND and (c) TD samples immersion in 3.5 wt% NaCl for 3h.

Fig.14.The surface morphology of the as-cast sample immersion in 3.5 wt%NaCl for 6h,presenting the typical galvanic corrosion.

Crystal orientation had important effects on corrosion resistance in Mg alloys due to the differences in surface energy.Generally,the (0001) crystal plane with higher atomic density (1.13×1019atoms/m2) compared with the (11-20) crystal plane(6.94×1018atoms/m2)and the(10-10)crystal plane(5.99×1018atoms/m2) for Mg alloys [52] had a lower electrochemical dissolution rate than the prismatic planes [53].The ND and TD samples exhibited a strong basal texture(Fig.6),in which the ND sample had (0001) oriented grains paralleled to surface,and prismatic planes were oriented towards surface for the TD sample.This further confirmed that the TD plane of the Mg-Zn-Ca-La alloy had a high hydrogen evolution rate and a low impedance value to the ND plane.Therefore,the ND sample with strong basal texture should be beneficial for the corrosion resistance improvement compared with the as-cast alloy with random orientations.Additionally,the grain size was another important factor for corrosion resistance for Mg alloys [46].It had been pointed out that grain boundaries were beneficial for the passivation effect enhancement [54],which was consistent with the dense protective film in ND sample at the initial stage of corrosion(Fig.12b3).Therefore,the grain size refinement after rolling deformation and annealing treatment should be responsible for the corrosion resistance improvement.

Fig.15 shows the schematic corrosion development of Mg-Zn-Ca-La alloy.At the initial stage of corrosion,a corrosion products layer,such as Mg(OH)2or MgCO3,was firstly formed on the surface of Mg alloys,which could prevent solution further penetrating into the matrix (Fig.15a1,b1and c1).The Mg(OH)2film formed on the surface of Mg alloys could protect the matrix from corrosion depending on the nature of the layer [35].The protective layer for the as-cast was porous and loose(Fig.12a3),while the ND sample contained a dense protective layer (Fig.12b3).Therefore,the surrounding solution could easily penetrate into the matrix of the as-cast alloy and react with the magnesium matrix.As a result,the as-cast alloy exhibited a more rapidly corrosion rate than the ND sample.Additionally,the ND sample with most of〈0001〉oriented grains parallel to the surface could provide extra protective effects for corrosion compared with the as-cast alloy with randomly orientations [55].As the corrosion proceeding,the protective layer was destroyed (Fig.15a2and b2).At this stage,the second phase and the matrix would form microgalvanic corrosion,in whichα-Mg had a lower corrosion potential than the second phases(Ca2Mg6Zn3and T).Therefore,Mg matrix was preferentially corroded.It should be noted that the prismatic planes of Mg alloys have a more negative potential than the basal planes.Consequently,high potential difference between 〈10-10〉 oriented grains of TD sample and second phase gave rise to rapid corrosion(Fig.15c2).Subsequent corrosion process was affected by the size and distribution of second phase [56].For the as-cast alloy,the network second phase provided a barrier for corrosion extending forward along the previously corroded positions,which resulted in corrosion preferentially expanding into grain interior (Fig.15a3).For the ND and TD sample,the second phase with fine distribution gave rise to the formation of more micro-galvanic sites.Consequently,more uniform corrosion morphology was observed.Additionally,the fine microstructure and(0001)oriented grains provided much more corrosion resistance for the ND sample (Fig.15b3).As a result,the ND sample exhibited a much slower corrosion rate.However,the galvanic couple with high potential difference of the TD sample promoted the corrosion expanding into the interior of matrix(Fig.15c3).

Fig.15.Schematic image of corrosion development for (a1)-(a3) the as-cast,(b1-b3) ND and (c1-c3) TD samples.

5.Conclusions

The corrosion behavior of a Mg-Zn-Ca-La alloy in the as-cast and as-rolled states immersed in 3.5wt% NaCl solution was investigated.The following conclusion points were drawn.

(1) Rolling deformation and subsequent annealing treatment improved the corrosion resistance significantly .

(2) The second phase with semi-continuous network of the as-cast alloy gave rise to the penetration of corrosion into the matrix.However,the second phase was dispersed after deformation,which resulted in uniform corrosion.

(3) The finel homogeneously microstructure was beneficial for the formation of dense corrosion layer at the initial stage of corrosion,which improved the corrosion resistance.

(4) The ND sample exhibited a good corrosion resistance,which was ascribed to the homogeneous microstructure and strong basal texture with 〈0001〉 oriented grains.

Declaration of Competing Interests

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.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (No.51801150),Shaanxi Natural Science Basic Research Program (No.2019JQ-512) and Shaanxi Provincial Department of Education Fund (No.16JK1557).