Effects of Microstructure on Discharge Behavior of AZ91 Alloyas Anode for Mg­Air Battery

Motohiro Yuasa1,+, Xinsheng Huang1, Kazutaka Suzuki1,Mamoru Mabuchi2 and Yasumasa Chino1

1Materials Research Institute for Sustainable Development, National Instituteof Advanced Industrial Science and Technology, Nagoya 463-8560, Japan2Department of Energy Science and Technology, Graduate School of Energy Science,Kyoto University, Kyoto 606-8501, Japan

The effects of precipitate distribution in commercial Mg alloys on the discharge behavior of a Mg­air battery were investigated. RolledMg­9mass%Al­1mass%Zn (AZ91) alloy sheets were selected as the anode. The discharge behaviors were evaluated by constant currentdischarge tests, and the corrosion behaviors were estimated by salt immersion tests and potentiodynamic polarization measurements. In thedischarge tests, the peak-aged specimens exhibited much shorter discharge time than the solution-treated specimen. In the corrosion tests, theaged specimens exhibited quite a low corrosion rate as compared with the solution-treated specimen. From the potentiodynamic polarizationmeasurements, the aged specimens showed a much lower corrosion current density and higher corrosion potential than the solution-treatedspecimen. It is suggested that a stable corrosion barrier by densely distributed ¢-phase precipitates contributed to both the reduction of dischargetime and the suppression of corrosion rate for the aged specimens. Microstructural evaluations of the aged specimens revealed that fine ¢-phaseprecipitates were densely distributed throughout the specimens and that there was little Al-rich ¡ phase, which would accelerate galvaniccorrosion between the ¡ matrix and the ¢ phase. It was confirmed that the finely and homogenously distributed ¢ phase in the aged AZ91specimens acted as a barrier to the dissolution of the ¡-Mg matrix because a large Al-rich ¡ phase was not distributed in the AZ91 specimens inthis study. [doi:10.2320/matertrans.MC201403]

(Received February 4, 2014; Accepted March 12, 2014; Published July 25, 2014)

Keywords: magnesium alloys, magnesium­air battery, corrosion, microstructure, heat treatment

1. Introduction

Mg alloys are promising light structural materials becauseof their low density, high specific strength and stiffness, andthey are widely used in the automobile and consumerelectronics industries.1) Recently, Mg alloys have alsoattracted attention as anodes in Mg­air batteries, which area promising power source and energy storage device owingto their high theoretical voltage, high current capacity, andenvironmental friendliness.2­4) However, Mg­air batteries arenot as well known as Al­ and Zn­air batteries because of theless attractive properties of Mg anodes. It is reported thateffective methods for improving the properties of Mg anodesinclude suppressing oxide film formation on the anodesurface and suppressing the excessive dissolution of theanodes during discharge.4­7) It is suggested that thecomposition of the Mg alloy anodes is one of the keyparameters for improving their properties. For example, Maet al.4) reported that Mg­Li­Al­Ce alloys exhibited higheranode performance than commercial Mg alloys (pure Mg andAZ31) because discharge products on the anode surfacebecame loose, which retained a large reaction surface duringdischarge. Zhao et al.7) reported the superior dischargebehavior of Mg­Ga­Hg alloy anodes as compared withanodes of commercial Mg alloys because the addition of Gaaccelerates stripping of the corrosion film and the formationof Mg amalgams, thus promoting electrochemical activity.

Meanwhile, differences in microstructure such as precip-itate distribution, are known to be important factors forcontrolling the corrosion behavior of Mg alloys.8­12) Con-cerning commercial Mg alloys (Mg­Al series alloys), the ¢

(Mg17Al12) phase provides a shield against corrosion of the¡ phase by forming a continuous network around the grainboundaries.13­16) On the other hand, the ¢ phase occasionallyacts as a cathode when it precipitates discontinuously in andaround grain boundaries.17­20) These reports indicate that amicrostructural factor may play an important role in thedischarge behavior of Mg alloy anodes; however, therelationship between the microstructure of Mg alloys andtheir discharge behavior as an anode is still unclear.

In the present study, the effects of precipitate distributionin commercial Mg alloys on the discharge behavior of aMg­air battery were investigated and the relationshipsbetween the microstructure, discharge behavior, and corro-sion behavior of the commercial Mg alloys as anodes arediscussed.

2. Experimental

AZ91 alloy sheets were selected as the Mg­air batteryanode in this study because precipitation (¢ phase)distribution is easily controlled by aging. The AZ91 alloysheets were prepared by the following procedure. An AZ91alloy bar 60mm in length, 50mm in width, and 5mm inthickness was extruded from a commercial AZ91 ingot. Thecomposition of the AZ91 alloy in this work is shown inTable 1. The extruded bar was heated at 703K and thenunidirectionally rolled at a rolling reduction of 20%. Theheating and rolling treatments were repeated until the alloyreached a thickness of 1mm. The rolled specimens were thensolution heat treated at 698K for 24 h. After that, aging wascarried out at 473K for 10, 16, 19, and 24 h.

The discharge performance of the Mg­air battery wasinvestigated by constant current discharge tests at a current+Corresponding author, E-mail: m-yuasa@aist.go.jp

Materials Transactions, Vol. 55, No. 8 (2014) pp. 1202 to 1207Special Issue on Platform Science and Technology for Advanced Magnesium Alloys, VI©2014 The Japan Institute of Metals and Materials

density of 10mA·cm¹2 at room temperature. Voltage versustime was recorded up to a cutoff voltage of 0.5V. The anodicefficiency and discharge capacity were calculated usingeqs. (1) and (2), respectively:

Anodic efficiency ð%Þ ¼ ði� A� tÞ �Ma

2FðWi �WfÞ� 100% ð1Þ

Discharge capacity ðmAh=gÞ ¼ i� A� t

Wi �Wf

ð2Þ

where i is the current density (mA·cm¹2), t is the dischargetime (h) of the discharge test, F is the Faraday constant(96485C·mol¹1), A is the surface area (cm2), Ma is the atomicmass (g·mol¹1) of the specimens, and Wi and Wf are theweight of the specimens before and after the discharge test,respectively.5)

The AZ91 specimens with dimensions of 25mm ©35mm © 1mm, an activated carbon sheet electrode (suppliedby the NIPPON VALQUA INDUSTRIES, LTD), and a5mass% NaCl aqueous solution were used as the anode,cathode, and electrolyte, respectively, of the Mg­air battery.The surfaces of the anode specimens were preliminarypolished with #1000-grit SiC abrasive paper.

For corrosion testing, AZ91 specimens prepared in thesame manner as the anode specimens were used. Saltimmersion tests were then performed for 72 h in a 5mass%NaCl solution saturated with Mg(OH)2, whose pH wasadjusted to 10.0. The samples were then exposed to ambientlaboratory temperature (308K). After testing, the corrosionproducts were stripped in a boiling chromic acid solutioncontaining 10mass% CrO3, and the corrosion rate wasdetermined from measurements of weight loss. Polarizationcurves were obtained under flowing N2 gas at roomtemperature in an electrolytic cell containing approximately

500 cm3 of the 5mass% NaCl solution saturated withMg(OH)2 (pH 10.0) using a Princeton applied researchVersaSTAT3 electrochemical measurement system. Thereference electrode was a saturated calomel electrode(SCE). Potentiodynamic polarization curves were measuredwith respect to the open-circuit potential at a scanning rate of1mV/s from ¹200 to 350mV.

The microstructures of the anode specimens were observedby optical microscopy (OM) and transmission electronmicroscopy (TEM). Phase identification of the anode speci-mens was performed by energy dispersive X-ray spectrosco-py (EDS) and X-ray diffractometry (XRD). The Vickershardness was measured at ten randomly selected points foreach sample using a load of 0.98N for 10 s.

3. Results and Discussion

Figure 1 shows the microstructures of the solution-treatedspecimen and the 10, 16, 19, and 24 h-aged specimens. Theaverage grain size of the solution-treated specimen and the10, 16, 19, and 24 h-aged specimens was 102, 106, 96, 99,and 100 µm, respectively, indicating that the grain size wasnot affected by the aging. The microstructure of the solution-treated specimen was composed of only a single phase (¡-Mgmatrix). In the case of the aged specimens, not only an ¡-phase region (white) but also a black region was found nearthe grain boundaries. Figure 2 shows the XRD patterns of allthe specimens in Fig. 1. In the aged specimens, a ¢-Mg17Al12phase was detected, in addition to the ¡ phase, implying thatthe black region in Fig. 1 corresponds to the ¢ phase. Itshould be noted that the peak intensity of the ¢ phase had amaximum value in the 19 h-aged specimen.

The age-hardening response of the AZ91 specimens at473K is illustrated in Fig. 3. The hardness increment of thespecimens was approximately 20HV, and the peak hardnesswas observed at 19 h. The aging time for peak hardnesscorresponded to the results of the XRD measurement, wherethe peak intensity of the ¢ phase had a maximum value at19 h.

Table 1 Chemical composition of AZ91 in mass percent.

Al Zn Mn Fe Cu Ni Mg

8.58 0.73 0.20 0.0021 0.0001 0.0004 Bal.

100μm 100μm 100μm

(a) (b) (c)

100μm

(d)

100μm

(e)

Fig. 1 Optical micrographs of the AZ91 alloys: (a) solution treated (ST) and (b) 10, (c) 16, (d) 19, and (e) 24 h-aged specimens.

Effects of Microstructure on Discharge Behavior of AZ91 Alloy as Anode for Mg­Air Battery 1203

The discharge behaviors of the Mg­air battery are shownin Fig. 4 and Table 2. Figure 4 shows the voltage­timecurves of the Mg­air battery, in which the solution-treated(ST) and aged specimens were applied as anodes, andTable 2 is the summary of the measured values (dischargetime, operating voltage, anodic efficiency, and dischargecapacity) from the voltage­time curves. In the solution-treated specimen, a constant voltage of 1.1V continued up to3 h and then decreased gradually. The final discharge timewas 6.56 h. However, the voltage rapidly decreased in allthe aged specimens as compared with the solution-treatedspecimen, and the 19 h-aged specimen exhibited the shortestdischarge time among the aged specimens. The 24 h-aged(over-aged) specimen had a slightly higher discharge timethan the 19 h-aged specimen. It seems that precipitation of thefine ¢ phase created by the aging process has a detrimentaleffect on the discharge time of the Mg­air battery. Largedifferences were not observed in the other measured values(operating voltage, anodic efficiency, and discharge capacity)of the solution-treated and aged specimens. Thus, it is clearthat the aging treatment of the AZ91 specimen affects the

discharge time of the Mg­air battery; however, it does notexert an influence on the other discharge properties.

Next, the corrosion and electrochemical properties of theAZ91 specimens were evaluated. The self-corrosion rateobtained by weight loss measurements is shown in Fig. 5.The corrosion rate of the solution-treated specimen wasapproximately 7 times higher than those of the agedspecimens. One researcher18) reported that a higher corrosionrate was obtained in a T6-treated cast AZ91 alloy ascompared with a solution-treated cast AZ91 alloy. On theother hand, other researchers14,21,22) reported that the T6-treated AZ91 cast alloy exhibited a lower corrosion rate thanthe solution-treated AZ91 cast alloy because finely and

MgMg17Al12Al6Mn

Inet

nsity

(a.

u.)

20 30 40 50 60 70 80

Diffraction angle, 2θ / deg

(a)

(b)

(c)

(d)

(e)

Fig. 2 XRD patterns of AZ91 specimens with (a) solution treatment and(b) 10, (c) 16, (d) 19, and (e) 24 h-aging treatment.

0 10 20 3060

70

80

90

100

Aging time, t / h

Har

dnes

s, H

/Hv

Fig. 3 Age-hardening response of AZ91 specimens at 473K.

ST

10h24h

16h

19h

0 2 4 60

0.5

1.0

1.5

Time, t /h

Vol

tage

. V /

V

Fig. 4 Voltage­time curves of the Mg­air battery, where solution-treated(ST) and aged AZ91 specimens were applied as the anode.

Table 2 Discharge performance of Mg­air battery with solution-treated(ST) and aged AZ91 specimens.

Heattreatment

Dischargetime(h)

Operatingvoltage(V)

Anodicefficiency

(%)

Dischargecapacity(mAh/g)

ST 6.56 0.95 54.6 1204

10 h-aging 5.31 0.74 56.4 1244

16 h-aging 1.15 0.91 55.2 1218

19 h-aging 0.66 0.82 52.9 1167

24 h-aging 3.48 0.80 55.7 1229

0

2

4

6

8

ST 10h 16h 19h 24h

Cor

rosi

on r

ate,

/mm

yea

r-1

Fig. 5 Corrosion rate of solution-treated (ST) and aged AZ91 specimensobtained from 5mass% NaCl salt immersion tests over 72 h.

M. Yuasa, X. Huang, K. Suzuki, M. Mabuchi and Y. Chino1204

homogenously distributed ¢ phases act as a barrier againstthe development of corrosion of the ¡ phase. The currentexperimental results seem to demonstrate a tendency towardthe latter results. It is considered that the low corrosion rate ofthe aged specimens in Fig. 5 is directly correlated with thesmall discharge time of the Mg­air cell with the agedspecimens as the anode, as shown in Fig. 4.

Figure 6 shows the potentiodynamic polarization curves ofthe solution-treated and aged specimens, and the corrosionparameters calculated by Tafel linear extrapolation aresummarized in Table 3. Explicit differences in the curves

were found between the solution-treated and aged specimens,and these results were well consistent with those of the self-corrosion rate measurements. The aged specimens exhibitedmuch lower corrosion current density than did the solution-treated specimen, and the 16 and 19 h-aged specimensshowed higher corrosion potential, which implies reducedelectrochemical activity.4,7) The sudden drop in the anodiccurrent density observed in the 10 and 24 h-aged specimensresults from local pitting of Mg and its alloys.14) Evidenceof local pitting confirms the destruction of oxide films orcorrosion products formed on the specimen surface. On theother hand, in the 16 and 19 h-aged specimens, a suddendrop in anodic current was not observed, indicating thatdissolution of the ¡ phase was difficult to accomplish, evenwhen an overpotential was applied. The polarization curvesindicate that a stable corrosion barrier was formed by denselydistributed ¢-phase precipitates in the peak-aged specimen,which caused both the reduction of discharge time and thesuppression of the corrosion rate.

Next, the reasons why the ¢ phase did not act as an anodeagainst the ¡ phase and instead behaved as a corrosion barrierare as follows. Figure 7 shows the transmission electronmicrographs of the 16 h-aged specimen. As shown in Fig. 1,the microstructures of the aged specimens were composedof a black region, where ¢-phase precipitates were denselydistributed, and a white ¡-phase region, where ¢ phaseprecipitates were not observed, at least at the magnificationof OM. On the other hand, fine, needle-like ¢ precipitatesapproximately 20 nm in width and a few hundred nanometersin length were observed in the ¡-phase region in Fig. 7. In theblack region, the ¢ phase precipitates as larger elliptical¢ precipitations than those in white region were denselydistributed. Figure 7 suggests that fine ¢-phase precipitateswere densely distributed throughout the aged specimens. Thehigh-magnification micrograph of the ¢-phase precipitatesand the results of the elemental analyses by EDS aresummarized in Fig. 8. The EDS results indicate the 16 h-agedspecimen had only a high concentration gradient with about20 nm at the precipitation boundaries. Hakamada et al.23)

Table 3 Corrosion parameters of solution-treated (ST) and aged AZ91specimens obtained from potentiodynamic polarization curves in Fig. 6.

Heat treatmentCorrosion potential

Ecorr (V/SCE)Corrosion current density,

lcorr (A·cm¹2)

ST ¹1.50 3.51 © 10¹5

10 h-aging ¹1.51 1.18 © 10¹6

16 h-aging ¹1.44 1.32 © 10¹6

19 h-aging ¹1.45 1.68 © 10¹6

24 h-aging ¹1.51 0.92 © 10¹6

10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100

-1.8

-1.6

-1.4

-1.2

Current density, i / A · cm-2

Pot

entia

l, E

/ V

SC

E ST

10h

24h16h

19h

Fig. 6 Polarization curves of solution-treated (ST) and aged AZ91specimens, where Epit indicates evidence of pitting corrosion.

200 nm 200 nm

(b)(a)

white region(α phase region)

black region(β phase region)

Fig. 7 TEM micrographs for 16 h-aged AZ91 specimen: (a) ¡-phase region in Fig. 1 and (b) boundaries between the ¡- and ¢-phaseregions shown in Fig. 1.

Effects of Microstructure on Discharge Behavior of AZ91 Alloy as Anode for Mg­Air Battery 1205

reported that the aluminum and zinc concentration gradientwith 4 µm thickness, which is much larger than that of foundin the present study, was distributed at the interface betweenthe secondary-phase precipitates and the ¡ matrix in anas-cast Mg­9mass%Al­1mass%Zn­1mass%Ca alloy. Thus,Fig. 8 indicates that there was minimal segregation of Al andZn solutes around the ¢-phase precipitates in the 16 h-agedspecimen. This minimal segregation is likely because atelevated temperatures, a large plastic deformation is imposedon the AZ91 specimens during sheet-making processes,which promotes diffusion of aluminum and zinc solutes intothe ¡ matrix.

Raman18) reported that the ¢ phase in the microstructure ofan as-cast AZ91 alloy is typically surrounded by an Al-rich¡ phase, which is defined as primary ¡ phase with higheraluminum solute concentration than ¡ matrix, and that thisphase is more anodic to the ¢ phase than to the primary¡ phase. He also reported that galvanic corrosion can beaccelerated with an increase in the ratio of cathode (¢ phase)to anode (Al-rich ¡ phase). On the other hand, as describedabove, other researchers8,21,22) reported that finely andhomogenously distributed ¢ phases in as-cast and agedAZ91 alloys act as a barrier against corrosion of the ¡

phase. In general, during solidification of AZ91 alloys, thealuminum solute is ejected into the liquid phase, leadingto continual enrichment of the liquid and lower soluteconcentrations in the primary ¡ phase. As a result, an Al-rich¡ phase tends to segregate around primary ¢-phaseprecipitates.24) It is known that the distribution of solutesegregation around a final solidification point stronglydepends on the solidification parameters such as the growthrate of the ¡ phase and the temperature gradient.25) Thisimplies that the distribution of the Al-rich ¡ phase in theAZ91 alloy is determined by process conditions duringcasting and that the role of the ¢ phase on the corrosionbehaviors may vary with the process conditions during

casting. As described above, the large area of the Al-rich ¡

phase was not detected in the aged AZ91 specimens in thisstudy because of the sheet making processes. Therefore, it isconclusively demonstrated that the finely and homogenouslydistributed ¢ phase in the aged AZ91 specimens acts as abarrier to the dissolution of the ¡-Mg matrix because thelarge area of Al-rich ¡ phase was not distributed in the AZ91specimens in this study.

It is interesting to note that the other discharge propertiessuch as operating voltage, anodic efficiency, and dischargecapacity of the aged AZ91 specimens had almost the samevalues as the solution-treated specimens, as shown inTable 2. Ma et al.4) reported the superior discharge properties(anode efficiency and discharge capacity) of the Mg­Li­Al­Ce alloy anode of a Mg­air battery, and suggested thatits superior discharge properties can be attributed to themodification of discharge products on the anode surface byalloying. The discharge products of the Mg­Li­Al­Ce alloybecome looser than those of AZ series alloys. These loosedischarge products promote many pores and cracks on thesurface film and contribute to the creation of a large reactionsurface during discharge, resulting in superior anodeefficiency and discharge capacity. A series of experimentsin this study and the results of a previous report7) indicate thatthe shape and distribution of ¢-phase precipitates in the anodespecimen likely do not affect the anode efficiency anddischarge capacity; however, the composition of the primary¡ phase likely plays a significant role in the dischargeproperties of Mg­air batteries. Investigations clarifying theeffects of alloy composition (aluminum and zinc concen-trations) on the discharge properties are still in progress.

4. Conclusions

The effects of precipitate distribution in AZ91 Mg alloyson the discharge behavior of a Mg­air battery wereinvestigated and the relationships between microstructure(precipitate distribution), discharge behavior, and corrosionbehavior were evaluated. The results are concluded asfollows:(1) In the discharge tests, the peak-aged specimens

exhibited much shorter discharge time than the solu-tion-treated specimen, whereas the other dischargeproperties such as operating voltage, anodic efficiency,and discharge capacity of the aged specimens hadalmost the same values as the solution-treated speci-men.

(2) In the corrosion tests, the aged specimens exhibited avery low corrosion rate as compared to the solution-treated specimen. From the measurements of potentio-dynamic polarization curves, the aged specimensshowed a much lower corrosion current density andhigher corrosion potential than the solution-treatedspecimen. Also, evidence of local pitting, whichindicates dissolution of the ¡ phase, was not observedin the peak-aged specimens. It is suggested that a stablecorrosion barrier of densely distributed ¢-phase precip-itates was formed in the peak-aged specimen and thatthe corrosion barrier resulted in both the reduction ofdischarge time and the suppression of corrosion rate.

Mg (at%) Al (at%) Zn (at%)1 65.9 29.1 5.02 87.6 10.4 1.93 93.1 5.4 1.54 93.3 4.9 1.7

1 2 3 4

20nm

Fig. 8 The magnified EDS image around the ¢ phase [shown as needle-likeprecipitates in Fig. 7(a)], in which the chemical compositions at theinterface between the ¢ and ¡-phases are listed.

M. Yuasa, X. Huang, K. Suzuki, M. Mabuchi and Y. Chino1206

(3) TEM observation and EDS analysis of the aged AZ91specimen revealed that fine ¢-phase precipitates weredensely distributed throughout the specimens and thatthere was little segregation of Al and Zn solutes aroundthe ¢-phase precipitates, which accelerated galvaniccorrosion between the ¡ and ¢ phases. Therefore, it isconfirmed that the finely and homogenously distributed¢ phase in the aged AZ91 specimens acted as barrier tothe dissolution of the ¡-Mg matrix because a large areaof Al-rich ¡ phase was not distributed in the AZ91specimens in this study.

REFERENCES

1) M. Hakamada, T. Furuta, Y. Chino, Y. Chen, H. Kusuda and M.Mabuchi: Energy 32 (2007) 1352­1360.

2) K. F. Blurton and A. F. Sammells: J. Power Sources 4 (1979) 263­279.3) L. Jörissen: J. Power Sources 155 (2006) 23­32.4) Y. Ma, N. Li, D. Li, M. Zhang and X. Huang: J. Power Sources 196

(2011) 2346­2350.5) D. X. Cao, L. Wu, G. L. Wang and Y. Z. Lv: J. Power Sources 177

(2008) 624­630.6) Y. Feng, R. C. Wang, K. Yu, C. Q. Peng, J. P. Zhang and C. Zhang:

J. Alloy. Compd. 473 (2009) 215­219.7) J. Zhao, K. Yu, Y. Hu, S. J. Li, X. Tan, F. W. Chen and Z. Yu:

Electrochim. Acta 56 (2011) 8224­8231.8) G. Song and A. Atrens: Adv. Eng. Mater. 1 (1999) 11­33.

9) H. Alves, U. Koster, E. Aghion and D. Eliezer: Mater. Technol. 16(2001) 110­126.

10) G. Song and A. Aterns: Adv. Eng. Mater. 5 (2003) 837­858.11) G. Song: Adv. Eng. Mater. 7 (2005) 563­586.12) M. C. Zhao, M. Liu, G. Song and A. Atrens: Corros. Sci. 50 (2008)

1939­1953.13) G. Song, A. Aterns and M. Dargusch: Corros. Sci. 41 (1999) 249­273.14) G. Song, A. L. Bowles and D. H. St John: Mater. Sci. Eng. A 366

(2004) 74­86.15) G. L. Song, A. Atrens, X. L. Wu and B. Zhang: Corros. Sci. 40 (1998)

1769­1791.16) A. Pardo, M. C. Merino, A. E. Coy, F. Viejo, R. Arrabal and S. Feliu,

Jr.: Electrochim. Acta 53 (2008) 7890­7902.17) M. Jönsson, D. Thierry and N. LeVozec: Corros. Sci. 48 (2006) 1193­

1208.18) R. K. S. Raman: Metall. Mater. Trans. A 35 (2004) 2525­2531.19) S. Mathieu, C. Rapin, J. Steinmetz and P. Steinmetz: Corros. Sci. 44

(2002) 2737­2756.20) S. Mathieu, C. Rapin, J. Hazan and P. Steinmetz: Corros. Sci. 45 (2003)

2741­2755.21) O. Lunder, J. E. Lein, T. K. Aune and K. Nisancioglu: Corrosion 45

(1989) 741­748.22) T. Beldjoudi, C. Fiaud and L. Robbiola: Corrosion 49 (1993) 738­745.23) M. Hakamada, A. Watazu, N. Saito and H. Iwasaki: Mater. Sci. Eng. A

527 (2010) 7143­7146.24) Y. Tamura, Y. Kida, A. Suzuki, H. Soda and A. McLean: Mater. Trans.

50 (2009) 579­587.25) D. J. Fisher and W. Kurz: Fundamentals of Solidification, (Trans Tech

Publications, Aedermannsdorf, 1989) pp. 63­92.

Effects of Microstructure on Discharge Behavior of AZ91 Alloy as Anode for Mg­Air Battery 1207