s,b,Ma

Magnesium

ermliq. Df theon o

2014 Published by Elsevier B.V.

matert castal use oalloys

added to MgAl alloys. For example, the morphologies of AlCeIMCs in MgAlRE alloys are reported [7,8]. Al11Ce3 and Al3Ceare acicular-like shaped and Al2Ce is quadrate-like shaped. Thethermal stability of Al11Ce3 and Al2Ce is investigated in

are a great num-IMCs in Musion coef573 K, 623

673 K are reported in literatures [10,14]. The intermetallicare investigated in the MgCe system through a diffusion[15]. In the diffusion couple, four intermetallics are formMg11Ce, Mg39Ce5, Mg3Ce and MgCe. Recently, diffusion couples ofMg0.667Ce0.333 and Al0.667Ce0.333 are employed to determine thethermodynamic evaluation and optimize the AlMgCe ternarysystem [16]. However, the interdiffusion of MgAlCe ternary sys-tem is not fully reported.

The present study employs the diffusion couple technique andfocuses on the IMCs formation at MgAl/MgCe interface. The

Corresponding author at: College of Materials Science and Engineering, NationalEngineering Research Center for Magnesium Alloys, Chongqing University, Chon-gqing 400044, China. Tel.: +86 135 94190166; fax: +86 65111140.

E-mail address: jiangbinrong@cqu.edu.cn (B. Jiang).

Journal of Alloys and Compounds 619 (2015) 411416

Contents lists availab

Journal of Alloys a

.e lthe formation of relatively thermally stable AlCe intermetalliccompounds (IMCs) [6].

The AlCe IMCs have attracted special attentions in recentyears. Many investigations of IMCs have been done when Ce is

studies of IMCs forming process. Previously, thereber of studies of the diffusion coefcients andthrough the diffusion couple technique. The diffof Al in Mg alloy and intermetallic phases athttp://dx.doi.org/10.1016/j.jallcom.2014.09.0710925-8388/ 2014 Published by Elsevier B.V.g alloycientsK andphasescoupleed as:castability and mechanical properties at room temperature [4].However, the application of commercially available MgAl basedalloys is limited to room or near room-temperature applicationsbecause of their inferior creep resistance and lower strength at ele-vated temperatures [5]. The addition of cerium (Ce) is one of theeffective methods to optimize the microstructure and improve ele-vated temperature mechanical properties of MgAl alloys, due to

to understand the complex phase transformations and precipita-tions behavior in materials. Although there have been active diffu-sion kinetics and thermodynamics studies on the MgAl [10,11]and AlCe [12,13] binary systems, no systematic thermodynamicsand diffusion kinetics studies have been performed for theMgAlCe ternary system so far.

The diffusion couple technique is valuably experimental for theIntermetallic compoundsDiffusionSubstitutionKinetics

1. Introduction

As the lightest metal structuralhave high specic strength, excellenbility [1,2]. The current commerciamainly focused on the MgAl-basedials, magnesium alloysbility, and easy recycla-f magnesium alloys is[3], as Al improves the

Mg3.4Al2.4Ce0.3Y0.3Mn alloy [9]. The alloy is annealed at473 K for 800 h, while A111Ce3 and Al2Ce phases exhibit good ther-mal stability with no signicant changes in size and shape. How-ever, there is still some scientic confusion about the phaseformation process in MgAlCe ternary system need to be claried.Knowledge of thermodynamics and diffusion kinetics is importantKeywords:on the basis of an effective Gibbs free energy model. The activation energy for the growth of the total dif-fusion reaction layer was 36.6 kJ/mol.The formation of intermetallic compoundof MgAl/MgCe diffusion couples

Jiahong Dai a, Bin Jiang a,b,, Xin Li a, Qingshan Yang aaCollege of Materials Science and Engineering, National Engineering Research Center forbChongqing Academy of Science and Technology, Chongqing 401123, ChinacNo. 59 Institute of China Ordnance Industry, Chongqing 400039, China

a r t i c l e i n f o

Article history:Received 24 July 2014Received in revised form 2 September 2014Accepted 7 September 2014Available online 16 September 2014

a b s t r a c t

The formation of AlCe intcouples prepared by solid48 h and 72 h, respectivelyand it substituted for Mg owere formed via the reacti

journal homepage: wwwduring interdiffusion

Hanwu Dong b, Xiangsheng Xia c, Fusheng Pan a,b

gnesium Alloys, Chongqing University, Chongqing 400044, China

etallic compounds (IMCs) during interdiffusion of MgAl/MgCe diffusionuid contact method was investigated at 623 K, 648 K and 673 K for 24 h,uring the whole diffusion process, Al was the dominant diffusing species,MgCe substrate. Five AlCe IMCs of Al4Ce, Al11Ce3, Al3Ce, Al2Ce and AlCef Al and Ce. The formation of Al4Ce as the rst kind of IMC was rationalized

le at ScienceDirect

nd Compounds

sevier .com/locate / ja lcom

MgCe substrate, as named as IIV, respectively. It can be found

Fig. 1. Optical micrographs of the diffusion reaction layers for the diffusion couplesannealed at (a) 623 K, (b) 648 K, and (c) 673 K for 72 h.

Fig. 2. X-ray diffraction patterns of the reaction products after annealed at 673 Kfor 72 h.

d CoIMCs formation is predicted by the thermodynamic model. Thegrowth kinetics of the diffusion reaction layers are determinedsimultaneously.

2. Experimental

A solidliquid contact method was employed to produce the MgAl/MgCe dif-fusion couples. Pure Mg (99.98 wt.%), pure Al (99.99 wt.%) and MgCe (Mg20 wt.%Ce, melting point: 868 K) master alloy were used. Al was added into Mg to form ofMgAl (Mg40 wt.% Al, melting point: 733 K) master alloys. The alloys were meltedseveral times to ensure the composition homogeneity. A rectangular piece of MgCemaster alloy was polished with 200-grit SiC paper, and then cleaned with acetone tomake sure an oxide free surface. MgAl master alloy was melted under a protectiveatmosphere of CO2 + 0.5 vol.% SF6. The melt was then taken from the furnace underthe cover gas of CO2 + 0.5 vol.% SF6 and MgCe master alloy was immediately sub-merged into the MgAl melt. Due to the higher thermal expansion coefcient ofMgAl alloy compared with that of MgCe alloy, an intimate contact betweenMgAl and MgCe alloys was formed during solidication. The samples wereencapsulated in quartz tubes. Quartz tubes were placed in a furnace, and heat trea-ted at 623 K, 648 K and 673 K for 24 h, 48 h and 72 h, respectively, and then thesamples were quenched in cold water to be rapidly cooled.

The diffusion couples were polished by conventional metallographic techniqueusing diamond paste with an oil-based extender and studied via optical microscopy(OM), scanning electron microscopy (SEM). The intermetallic phases formed at thediffusion reaction layers were identied by energy dispersive spectroscopy (EDS)and X-ray diffractometer (XRD). To determine the growth kinetics of the diffusionreaction layers, the thicknesses of the diffusion reaction layers were examined fromoptical micrographs of the cross-sections. An average thickness was obtained from20 individual measurements.

3. Results and discussion

3.1. Identication of reaction products

Optical micrographs of the cross-sections of the diffusion cou-ples annealed at 623 K, 648 K and 673 K for 72 h are shown inFig. 1. As is seen in these micrographs, several reaction sub-layersare observed to form at the MgAl/MgCe interfaces, and locatedin the MgCe substrate. The thickness of the total diffusionreaction layer increased evidently when increasing the annealingtemperature, which is about 155 lm, 173 lm and 200 lm inFig. 1(a)(c), respectively. It can be seen that signicant micro-structure variations are found for the different temperatures.

In order to identify the phase constitution of the diffusion reac-tion layers, X-ray diffraction examination is carried out. The X-raydiffraction on the diffusion reaction layers shown in Fig. 1(c) is car-ried out on two sub-layer after grinding down to the correspondingdepth. Indexing the spectra shown in Fig. 2, Al4Ce, Al11Ce3, Al3Ceand Mg are identied on the lower sub-layer. Different to the lowersub-layer, there are Al2Ce and AlCe in the top sub-layer. This isattributed to the complexity of these phases mixture of suchregions. Further investigations are required to elucidate the forma-tion of the IMCs microstructures.

3.2. Microstructural of the diffusion reaction layers

Fig. 3(a)(c) shows SEM back-scattered electron (BSE) images ofthe different layers formed in the diffusion couples annealed at673 K for 24 h, 48 h and 72 h, respectively. It is observed that sev-eral reaction layers exist between the two substrates. The micro-structure varies for different holding times and the totalthickness of these layers increases signicantly. At the holdingtime of 24 h, the acicular-like intermetallics in the reaction layerare not obvious. When the holding time reaches 48 h, theacicular-like intermetallics can be obviously seen in the reactionlayer. With further increasing of holding time, the thickness ofthe acicular-like intermetallics layer increases to about 110 lm.

412 J. Dai et al. / Journal of Alloys anIn order to study the diffusion behavior thoroughly, Fig. 3(c),which is for the sample annealed at 673 K for 72 h, is taken intoconsideration. Four layers are visible from the MgAl substrate tompounds 619 (2015) 411416that there are kinds of IMCs with different shapes in the diffusionreaction layers. The existence of diffusion reaction layer I may be aresult of Als intrinsically faster diffusion. In this layer,

d Compounds 619 (2015) 411416 413J. Dai et al. / Journal of Alloys anconcentration of Al lower than the substrate of MgAl, and it pre-sents a distribution of gradient decrease distribution showed Fig. 4.

Fig. 3(d)(f) is high magnication images of the regions 13,respectively. For layers II and III, they are mainly composed of par-ticle-like intermetallics, but the macro morphology is different. Inthe layer IV, there are a lot of acicular-like intermetallics, which aregrown along the diffusion direction. Besides, a few quadrate-likeintermetallics particles can be found by careful BSE observationsnearby the MgCe substrate. While the EDS results taken fromthe testing positions denoted in Fig. 3(d)(f) are summarized inTable 1. Nevertheless, AlCe is not detected in BSE micrograph, evenafter careful examination along the interface by using BSE at highmagnication. The probable reason is that the AlCe phase is toosmall to be resolved by the used characterization techniques.

Fig. 4 shows the analysis of the EDS line scan across the diffu-sion couple in the sample is annealed at 673 K for 72 h, shown in

Fig. 3. SEM images of the diffusion reaction layers for the diffusion couples annealed at 673 K for (a) 24 h, (b) 48 h, (c) 72 h, and high magnication images of (d) the region 1,(e) the region 2, (f) the region 3.

Fig. 4. Composition prole along the line scan shown in Fig. 3(c).

Fig. 3(c). The line scan contains data for 20 points. The concentra-tion prole of Mg is a slight uctuation in layers II and III, but it isincreased in layers I and IV. The distribution of Al is decreasedgradually in layers I and IV, while the Al content remains to be aconstant from layer II to layer III. The concentration prole of Ceis almost at in all zones. Because of the higher Al content in theMgAl substrate and the higher Mg content in the MgCe sub-strate, Al atoms incline to diffuse to MgCe substrate and Mgatoms to MgAl substrate. The concentration proles show thatthe substitution of Mg is taken place by Al, and AlCe IMCs areformed by AlCe reaction, along with their diffusion direction. It

section at 673 K, it can be concluded that Al11Ce3, Al3Ce and Al2Ceare identied in the diffusion reaction layers.

3.3. Formation of intermetallic compounds

Annealing in the temperature range of 623673 K in the presentstudy resulted in the formation of Al4Ce, Al11Ce3, Al3Ce, Al2Ce andAlCe between MgAl/MgCe in the diffusion process. The thermo-dynamic behavior is an important factor for the sequence of AlCeIMCs layer. Since MgAl, MgCe and AlCe compounds can form inthe MgAlCe ternary system. The Miedemas model [18] for theenthalpy of formation is used to determine the standard molarenthalpies of the formation of MgAl, MgCe and AlCe binary sys-tems. As shown in Fig. 6(a), the standard molar enthalpy of the for-mation of the AlCe binary system is far lower than those of MgCeand MgAl binary systems, which indicates that AlCe IMCs preferto be formed in the MgAlCe ternary system in the samecondition.

In previous studies [12,19,20] showed that the thermochemicalproperties of AlCe phases have been studied, but the values of thestandard enthalpy of the formation of these phases are widely dis-crepant. So the effective heat of formation (EHF) model cannot beused to predict the formation of the rst phase in AlCe binary sys-tem. It is well known that the Gibbs free energy of a chemical reac-tion is an effective criterion to judge whether or not this reactionwill occur. If Gibbs free energy is less than zero, the chemical reac-tion occurs. Moreover, among all possible reactions, the reactionwith the lowest Gibbs free energy tends to occur.

The Gibbs free energy for a solid solution phase in a binarysystem AB can be represented mathematically as follows [21]:

Table 1EDS results taken from different positions of the diffusion layers as denoted in Fig. 3.

Zone Mg (at.%) Al (at.%) Ce (at.%) Al/Ce ratio Phase

A 68.61 25.57 5.82 4.39 Al4CeB 73.82 21.10 5.08 4.15 Al4CeC 67.18 25.98 6.84 3.79 Al11Ce3D 63.97 27.82 8.21 3.38 Al3CeE 75.99 15.90 8.11 1.96 Al2CeF 94.93 5.07 0 Solid solution

414 J. Dai et al. / Journal of Alloys and Compounds 619 (2015) 411416can be concluded that the total diffusion reaction layers locatedin the MgCe substrate, which is in good agreement with opticalmicrographs in Fig. 1. This suggests that Al is the dominant diffus-ing species during the diffusion process, which may be related thatthe diffusion of Al is faster than that of Ce in Mg. Al has a lowermelting point (933 K) than Ce (1072 K), and it may require lessactivation energy for Al to migrate. In addition, the atomic radiusof Al (143 pm) is smaller than Ce (185 pm) which can allow Al tomove easily than Ce in Mg.

Fig. 5 shows the MgAlCe calculated isothermal section at673 K [17], where the points are the line scan of the diffusioncouple annealed at 673 K for 72 h, and mainly located inAl11Ce3 + (Mg), Al11Ce3 + Al3Ce + (Mg), Al3Ce + Al2Ce + (Mg) andAl2Ce + (Mg) phase elds. The diffusion path of the diffusion coupleis along the line with points. Nevertheless, no related data for Al4Cein ternary phase diagram of the MgAlCe system was reported, soit cannot nd the corresponding region in the phase diagram. AlCecannot be determined in the phase elds of the diffusion paththrough. From the analysis of the MgAlCe calculated isothermalFig. 5. MgAlCe calculated isothermal section at 673 K.Fig. 6. (a) Formation enthalpies of the MgAl, MgCe and AlCe binary systems and(b) Gibbs free energies of the formation of different AlCe IMCs as the function oftemperature.

K K0 exp Q0RT

6

or

lnK lnK0 Q0RT 7

where K0 is the pre-exponential factor, Q0 is the activation energy, Ris the gas constant, and T is the absolute temperature. The activa-tion energy can be obtained from the plots of lnK versus the reci-procal of temperature T, as shown in Fig. 7(b). The calculatedapparent activation energy for the growth of the total diffusionreaction layer is 36.6 kJ/mol. The activation energy is slightly lowerthan those of Brennan et al. [10] and DAS et al. [11], which may berelated to the melting temperature of the MgAl alloy is lower thanAl [24].

4. Conclusions

The formation of intermetallic compounds (IMCs) during inter-diffusion of MgAl/MgCe diffusion couples has been investigatedat 623 K, 648 K and 673 K for 24 h, 48 h and 72 h, respectively. Theconclusions are summarized as follows:

(1) During the whole diffusion process, Al was the dominant dif-fusing species. Al atoms substituted for Mg ones and formedAlCe IMCs.

(2) Five AlCe IMCs of Al4Ce, Al11Ce3, Al3Ce, Al2Ce and AlCe were

Fig. 7. (a) Relationship between the thickness of the diffusion reaction layers andthe square root of treatment times for all the diffusion couples and (b) an Arrheniusplot of the diffusion reaction layers growth.

d CoGS CAGA CBGB DHSmix TDSSmix 1where CA and CB are the mole fractions of A and B elements, respec-tively. GA and GB are the Gibbs free energy of the pure element A andB, respectively. DHSmix is the enthalpy change during intermixing,DSSmix RCA ln CA CBlnCB is the ideal entropy of intermixing,and T is the temperature. The enthalpy change of a solid solutionphase can be determined by the semi-experimental Miedemasmodel:

DHSmix DHC DHE DHS 2where the three terms are the chemical, elastic, and structural con-tributions, respectively. The chemical term DHC can be expressedby:

DHC CACB CSBDHSolA in B CSADHSolB in A

3

where DHSolA in B is the solution enthalpy of A in B, CSB is the degree to

which A atoms are in contact with B atoms. The elastic term for asolid solution phase is often calculated by:

DHE CACB CBDHeA in B CADHeB in A 4

where DHeB in A is the elastic contribution to the heat of solution of Bin A in the solid solution phase. DHS is different from zero only fortransition metaltransition metal alloys [22]. For the AlCe solu-tions, Al a main group metal and Ce is a transition metal, so thestructure contribution DHS 0.

Based on the equations described above, the values of Gibbs freeenergies are calculated in the temperature range 623673 K andthe results obtained are shown in Fig. 6(b). The parameter valuesfor calculations are taken from the literature [23]. It can be foundin this gure that in this temperature range, the Gibbs free energyof Al4Ce is more negative than the other phases, resulting in thepreferential formation of Al4Ce at the interface, and Al11Ce3 is thesecond, and the sequence of following IMCs is Al3Ce, Al2Ce andAlCe, which are in good agreement with the experimentalobservations.

3.4. Growth behavior of the diffusion reaction layers

Growth kinetics for the diffusion reaction layers are evaluatedby examining all the diffusion couple samples. The experimentalgrowth rates of the diffusion reaction layers follow typical parabolictrends with time, suggesting that the growth is controlled by diffu-sion process, and can be described by the empirical equation:

d Kt1=2 5where d is the thickness of the diffusion reaction layer in meter andt is the reaction time in second, and K the growth constant in m2/s.The growth constant K was dened in such a way so that it had thesame dimension as the diffusion coefcient.

The growth kinetics for the diffusion reaction layers are pre-sented in Fig. 7(a). It can be seen that linear regression analysisgives best-t straight lines, and most of the linear correlation coef-cient values (R2) for these plots are higher than 0.98. The resultindicated that the growth kinetics is well conformed to the para-bolic law, so the growth of the diffusion reaction layers is volumediffusion-controlled in the studied temperature range [21]. WithEq. (5) the reaction rates of the diffusion reaction layers formationare calculated. At 623 K, 648 K and 673 K, the growth constants ofthe total diffusion reaction layers are 8.90 1014 m2/s,1.15 1013 m2/s and 1.50 1013 m2/s, respectively. These K val-ues are similar to the experimental ones of Brennan et al. [10],

J. Dai et al. / Journal of Alloys anespecially at 623 K.The following Arrhenius-type relationship is used to determine

the apparent activation energy for diffusion controlled growth:mpounds 619 (2015) 411416 415formed via the reaction of Al and Ce. The deduction of ther-modynamics explains the formation of Al4Ce as the rstphase in the diffusion reaction interface.

(3) The growth of the IMCs was volume diffusion-controlled. At623 K, 648 K and 673 K, the growth constants of the totaldiffusion reaction layers were 8.90 1014 m2/s,1.15 1013 m2/s and 1.50 1013 m2/s, respectively. Theactivation energy for the growth of the total diffusion reac-tion layer was calculated to be 36.6 kJ/mol.

Acknowledgments

Theauthors are grateful for thenancial supports fromChongqingScience and Technology Commission (CSTC2013jcyjC60001,CSTC2012JJJQ50001, cstc2012ggB50003), National Natural ScienceFoundation of China (51171212), and The National Science andTechnology Program of China (2013DFA71070, 2013CB632200),and the Fundamental Research Funds for the Central Universities(CDJZR13138801, CDJXS12131106).

References

[1] B.L. Mordike, T. Ebert, Mater. Sci. Eng. A 302 (2001) 3745.[2] A.A. Luo, J. Magnes. Alloys 1 (2013) 222.[3] T. Zhu, P. Fu, L. Peng, X. Hu, S. Zhu, W. Ding, J. Magnes. Alloys 2 (2014) 2735.[4] J. Bai, Y. Sun, F. Xue, S. Xue, J. Qiang, T. Zhu, J. Alloys Comp. 437 (2007) 247

253.

[5] J. Zhang, J. Wang, X. Qiu, D. Zhang, Z. Tian, X. Niu, D. Tang, J. Meng, J. AlloysComp. 464 (2008) 556564.

[6] T.L. Chia, M. Easton, S.M. Zhu, M. Gibson, N. Birbilis, J. Nie, Intermetallics 17(2009) 481490.

[7] X. Tian, L. Wang, J. Wang, Y. Liu, J. An, Z. Cao, J. Alloys Comp. 465 (2008) 412416.

[8] T. Rzychon, A. Kiebus, J. Cwajna, J. Mizera, Mater. Charact. 60 (2009) 11071113.

[9] J. Zhang, Z. Leng, S. Liu, M. Zhang, J. Meng, R. Wu, J. Alloys Comp. 509 (2011)L187L193.

[10] S. Brennan, K. Bermudez, N.S. Kulkarni, Y. Sohn, Metall. Mater. Trans. A 43(2012) 40434052.

[11] S.K. Das, Y.M. Kim, T.K. Ha, R. Gauvin, I.H. Jung, Metall. Mater. Trans. A 44(2013) 25392547.

[12] Y.B. Kang, A.D. Pelton, P. Chartrand, C.D. Fuerst, Calphad 32 (2008) 413422.[13] L. Jin, Y.B. Kang, P. Chartrand, C.D. Fuerst, Calphad 35 (2011) 3041.[14] S. Brennan, A.P. Warren, K.R. Coffey, N. Kulkarni, P. Todd, M. Kilmov, Y. Sohn, J.

Phase Equilib. Diffus. 33 (2012) 121125.[15] X. Zhang, D. Kevorkov, M.O. Pekguleryuz, J. Alloys Comp. 501 (2010) 366370.[16] L. Jin, D. Kevorkov, M. Medraj, P. Chartrand, J. Chem. Thermodyn. 58 (2013)

166195.[17] V. Raghavan, J. Phase Equilib. Diffus. 28 (2007) 453455.[18] A.R. Miedema, P.F. Dechatel, F.R. Deboer, Physica B 100 (1980) 128.[19] G. Borzone, G. Cacciamani, R. Ferro, Metall. Mater. Trans. A 22 (1991) 2119

2123.[20] G. Cacciamani, R. Ferro, Calphad 25 (2001) 583597.[21] Y. Guo, G. Liu, H. Jin, Z. Shi, G. Qiao, J. Mater. Sci. 46 (2010) 24672473.[22] H. Bakker, G. Zhou, H. Yang, Prog. Mater. Sci. 39 (1995) 159241.[23] A.T. Dinsdale, Calphad 15 (1991) 317425.[24] J. Hirmke, M.X. Zhang, D.H. St John, Metall. Mater. Trans. A 43 (2012) 1621

1628.

416 J. Dai et al. / Journal of Alloys and Compounds 619 (2015) 411416