Effect of heat treatment on the intermetallic layer of cold sprayed aluminum coatings on magnesium alloy

Surface & Coatings Technology 205 (2011) 4665–4671

Contents lists available at ScienceDirect

Surface & Coatings Technology

j ourna l homepage: www.e lsev ie r.com/ locate /sur fcoat

Effect of heat treatment on the intermetallic layer of cold sprayed aluminum coatingson magnesium alloy

Hengyong Bu a,⁎, Mohammed Yandouzi b, Chen Lu a, Bertrand Jodoin b

a National Engineering Research Center of Light Alloy Net Forming, School of Materials Science and Engineering, Shanghai JiaoTong University, Shanghai 200240, PR Chinab Mechanical Engineering Department, University of Ottawa, Ottawa, K1N 6N5 ON, Canada

⁎ Corresponding author. Tel.: +86 21 54742618; fax:E-mail address: [email protected] (H. Bu).

0257-8972/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.surfcoat.2011.04.018

a b s t r a c t

Article history:Received 23 January 2011Accepted in revised form 3 April 2011Available online 9 April 2011

Keywords:Aluminum coatingsMagnesium alloysHeat treatmentCold sprayIntermetallic phases

Dense and thick pure aluminum coatings were deposited on AZ91D-T4 magnesium substrates using the coldspray process. Heat treatments of the as-sprayed samples were carried out at 400 °C using different holdingtimes. The feedstock powder, substrate and coating microstructures were examined using optical microscopy(OM), scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS) as well as Vickersmicrohardness analysis. The results demonstrate that aluminum coatings having dense and uniformmicrostructure can be deposited successfully using a relatively large feedstock powder. It has been identifiedthat the intermetallics Al3Mg2 (γ phase) and Mg17Al12 (β phase) were formed at the coating/substrateinterface during heat treatment. The growth rate of these intermetallics follows the parabolic law and the γphase has a higher growth rate than the β phase. The thickness of the Mg17Al12 and Al3Mg2 intermetalliclayers has reached 83 μm and 149 μm, respectively. This result is almost 45% higher than what has beenreported in the literature so far. This is attributed to the fact that T4 instead of as cast Mg alloy was used assubstrate. In the T4 state, the Al concentration in the Mg matrix is higher, and thus intermetallic growth isfaster as less enrichment is required to reach the critical level for intermetallic formation in the substrate. TheAZ91D-T4 magnesium substrate contains single α phase with fine clusters/GP-zones which is consideredbeneficial for the intermetallic formation as well as the intimate contact between the coating/substrateinterface and the deformed particles within the coating.

+86 21 34202794.

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Magnesium (Mg) alloys are attractive materials for a wide numberof applications owing to their suitable mechanical properties,low density, as well as their vast reserve [1]. In addition, their highspecific strength, good damping capacity and recyclability make themsuitable solutions for the automotive, aerospace and electronicindustries [1–3]. However, Mg alloys are susceptible to react withvarious metals or ions (such as chloride ions). Consequently, theyhave poor chemical and galvanic corrosion performance [1,4,5].Furthermore, inevitable impact or scratch of Mg components willlead to significant mass loss due to its low hardness [6]. These poorproperties increase the maintenance costs and limit the life cycle ofMg alloy components. One way to prevent these defects is tomanufacture a coating which has better performance and isolate theMg alloymatrix from the environment, without negative effects to theMg-base alloy [7,8]. Several coating techniques such as anodizingvapor deposition, electroplating and chromate coating, to name a few,

have been used for magnesium protection. Nevertheless, they exhibitsome limitations, such as low deposition efficiency, high powerrequirement or being environment unfriendly [8].

Cold spray is a coating technology which uses high pressure gas toaccelerate solid powder particles beyond a critical velocity [9–13]. Uponimpact with the substrate, the particles undergo severe plastic defor-mation and form a coating. Compared to other coating manufacturingmethods such as thermal spray or electroplating, cold spray hasdemonstrated many advantages with the most important being thatthe coatings retain the originalmicrostructure of the feedstock powders[10,11,14]. Using the cold spray process, previous studies have reportedthe possibility of protectingMg-alloy using aluminum (Al) andAl-basedalloy coatings [2,6,7,15–17]. It was found that Al coatings on ZE41A-T5substrates have similar performances than bulk Al and can improve thecorrosion resistance compared to the uncoated substrate [2,6].

In order to attain higher bonding strength and promote the overallcoating performance, cold spray coatings have been heat treated. Thishas resulted in element diffusion to form intermetallic compoundswithin Al/Ni [18], Al/Ti [10,19] and Al/Mg [15,20] couples. Annealedcold spray Al coatings on magnesium substrates have revealed theformation of the intermetallic β and γ layers near the substrate/coating interface. It has been reported that the later phases havesimilar corrosion resistance than Al alloys [15,20].

4666 H. Bu et al. / Surface & Coatings Technology 205 (2011) 4665–4671

The advantage of generating intermetallic compounds using coldspray combined with post heat treatment can be summarized asfollows: First, it reduces shrinkage during subsequent annealing if thecoatings have a high density, which is different from conventionalsintering that exhibits large porosities [21]. Second, the formedintermetallic compounds are continuous and uniform onto theoriginal substrate and exhibit higher hardness and better corrosionresistance [15].

Different intermetallic phases with different thicknesses havebeen reported for Mg/Al interdiffusion process during heat treatment.Shigematsu et al. [4] reported the formation of the Mg17Al12 phasewith approximately 750 μm layer thickness after 1 h at 450 °C usingpowder metallurgy technology. Using pure Al powders with a meandiameter of 15 μmas feedstock in the cold spray process, Spencer et al.[15,20] have identified the existence of two intermetallic phases(Mg17Al12 and Al3Mg2) at the Mg/Al interface with a total thicknessaround 200 μm after 24 h at 400 °C. Liu et al. [22] have identified thepresence of three intermetallic phases (MgAl, Mg3Al2 and Al3Mg2)during the vacuum diffusion bonding of the joint Mg and Al bulkmaterials annealing at 480 °C for 1 h. Consequently, it can beconcluded that the type and thickness of intermetallic phases dependon the heat treatment conditions.

Large feedstock powders have attracted much attention from thecold spray community as they have lower critical velocity and decreasedthe risk of explosion compared to fine powders [14,23–25]. As such,there is a considerable interest to determine and evaluate theintermetallic compounds fabricated by cold spray using relative largepowders and subsequent heat treatment. The aim of this work is to usethe cold spray process to spray Al ontoMg alloy using large Al feedstockpowders. AZ91D-T4magnesiumsubstratewas selected since it containsa uniform solid solution α-phase with clusters/GP-zones and thealuminum concentration in the magnesium matrix is higher. Thereforeintermetallic growth is faster because less enrichment is required toreach the critical level for intermetallic formation in the substrate. Theeffects of different heat treatments on the type of intermetallic phasesand their thicknesses are evaluated and discussed.

2. Experimental procedures

The coatings were produced at the University of Ottawa Cold SprayLaboratory. Details of the spray system can be found elsewhere[26,27]. Helium was used as both propellant gas and powder carriergas, although nitrogen can be used if operating at higher pressure. Thespraying parameters used in the currentwork are listed in Table 1. Thefeedstock powders were accelerated using a De Laval nozzle whichhas an expansion ratio of 10. Pure commercially available aluminumpowder (Al-101, Centerline (Windsor) Ltd, Canada) was used asfeedstock. The powder size distribution was measured by a laserparticle size analyzer (Coulter LS, Fullerton, CA). Commerciallyavailable AZ91D-T4magnesium substrates produced by High PressureDie Casting (HPDC)were used after being submitted to a solution heattreatment at 400 °C for 12 h and followed by quenching in water.Before spraying the Al powders, the natural aging duration to T4 statesubstrates was no longer than 48 h and substrates with dimensions of170×20×6 mmwere grit blasted using (20 mesh – 24 grit) silica andcleaned using acetone. The substrates were then assembled on a two-axis computer controlled moving system.

Table 1Parameters for cold spray process used for spraying pure aluminum (Al-101) powders.

Gas parameters Spray distance(mm)

Numberof pass

Pressure (MPa) Temperature (°C)

0.98 300 10 1

The coated samples were sectioned and prepared for microscopyanalysis, following standard metallographic methods. The etchingsolutions for magnesium substrate and Al coating were 4% nitric acid+96% ethanol and Kroll's reagent, respectively. The feedstock powder andcoating microstructures were characterized using Optical Microscopy(OM) and Scanning Electron Microscopy (Zeiss EVO-MA10, LaB6Analytical SEM) equipped with an Energy Disperse Spectroscopy(INCA X-act, Oxford, UK). Phase identification was investigated byX-ray diffraction (XRD) using a Philips X-Pert model 1830 X-raydiffractometer equippedwith a graphitemonochromator using Cu kα(λ=0.15406 nm) radiation. Detailed scans were performed over a20–100° 2θ range, 0.02° step width and 2 s per step acquisition time.Images of the cross sections were used to evaluate the coatingporosity level using the commercially available Clemex imagingsoftware. The powder and the coating microhardness were evaluatedusing a load of 25 g (HV0.025). All measurements used a dwell time of15 s, usingDuramin-2 Vickers hardness tester (Struers Inc., Cleveland,OH, USA). The reported hardness values are averages of at least 6random measurements for each sample. The heat treatments of theas-sprayed coatings were carried out in a vacuum furnace underpressure of 5.0×10−4 bars. The samples were heated at constant heattreatment temperature (400 °C) for different period of time up to 20 hwith a temperature fluctuation below 1 °C.

3. Results and discussion

3.1. Characteristics of pure Al powder

The morphology of the Al-101 feedstock powder is shown inFig. 1a. SEM observations of the free standing particles reveal theirregular morphology of the particles varying from spherical toelongated structures with different sizes. Fig. 1b presents an OMimage of the polished cross section powder. The image shows adendritic microstructure within the Al particle, caused by the rapidsolidification rate experienced by the particle during the powderatomization process. The particle size distribution diagram obtainedusing laser size analyzer is shown in Fig. 1c. It indicates that thepowder particles present a broad size distribution with a diameterranging from 6 to 174 μm, and the particles larger than 50 μmrepresent approximately 40% of the particles.

3.2. Properties of the as sprayed coatings and substrate

3.2.1. SubstrateThe microstructure of the high pressure die casting AZ91D

magnesium alloy is shown in Fig. 2a. It contains the primary αphase (Mg) matrix as well as the discontinuous intermetallic β phase(Mg17Al12) [28]. The later precipitates are located mostly at the αphase dendrite grain boundaries. After T4 heat treatment, theprecipitates (β phase) were dissolved into the matrix and themicrostructure of the Mg alloy displays a single Al supersaturated αphase as seen in Fig. 2b. Note that due to small size of clusters/GPzones that were formed inside the Al supersaturated grains during thenatural aging, they could not be seen by using OM or SEM imagingtechnique. The XRD results of the as cast and T4 magnesiumsubstrates are shown in Fig. 3, which also demonstrate that aftersolution treatment, the β phase peaks have disappeared.

3.2.2. Coating characteristicsAn SEM image under secondary electron mode of the as-sprayed

coating cross section is shown in Fig. 4. The coating has a thickness ofapproximately 300 μm and limited porosity (less than 1%). The overallcoating is dense with the exception of the top coating surfacewhere limited porosity was observed. This is attributed to the factthat the large particles do not undergo enough plastic deformationupon impact and thus leave some porosity at the particle–particle

Fig. 1. Al-101 feedstock powder: (a) SEM image shows the morphology of the freestanding particles. (b) OM image of resin embedded powder cross section afterpolishing and weak etching revealing the dendrite microstructure of the Al particle.(c) Size distribution of the Al powder measured by laser size analyzer.

Fig. 2. OM image showing (a) the microstructure of high pressure die casting AZ91Dmagnesium solid solution alloy α phase matrix and the β phase precipitates. (b) AZ91Dmagnesium alloy after T4 heat treatment (400 °C×12 h and quenched in water).

Fig. 3. XRD results of the as cast and T4 treated AZ91D magnesium substrates.

4667H. Bu et al. / Surface & Coatings Technology 205 (2011) 4665–4671

boundaries. In addition, the absence of significant peening effect thatsometimes contributes to extra deformation of the particles alreadydeposited leads to the increased porosity at the top of the coating. Thisphenomenon has already been described in other works [29]. Theinterface between the Al coating and the Mg substrate has almost noporosity, which means the particles near the interface had enoughvelocity to remove the surface oxide to expose fresh material,resulting in the intimate contact at the interface. This is consideredbeneficial for coating bonding strength and for enhanced diffusion ofthe elements during the subsequent heat treatment.

The feedstock powder used in this study presents a large sizedistribution, ranging from less than 10 μm to over 170 μm. It is knownthat large powder particle size distributions have a detrimental effecton the coating quality because the large particles are harder toaccelerate than the smaller oneswhile the lattermay be affected by thebow shockwave present in front of the substrate [30]. Consequently, itis difficult for all of the particles to reach a similar high velocity, andthus it can be expected that the large feedstock powder particle sizedistribution used in this work will typically result in a large velocitydistribution. As such, the particleswhichhave a velocity lower than thecritical velocity will rebound from the substrate surface or adhere to

the substrate with a low bonding strength andminimum deformationlevel, increasing the possibility of forming porosities. The morphologyof theAl particles in the coating after etching is shown in Fig. 5. It can beseen that there were several large particles in the coating which werelabeled by black arrow if their size was apparently larger than 50 μm,and the total area of the larger particles found in the coating wasestimated to be 42%. Compared to the ratio of the particles larger than50 μm in the feedstock powder, it is clear that the deposition efficiencyis similar for all particle sizes in the current work. Since the producedcoatings exhibit low porosity level, it is concluded that the disadvan-tages of large powder size distributionwere alleviated by the choice ofoptimized spraying parameters.

Fig. 4. SEM images of the cross section of Al coatings onto AZ91D magnesium.

Fig. 6. Al–Mg binary phase diagrams.

4668 H. Bu et al. / Surface & Coatings Technology 205 (2011) 4665–4671

Compared to themicrohardness of Al feedstock powder of 39.6±2.4HV0.025, the hardness of the as sprayed coatingwas increased to 60±3.4HV0.025. It is concluded that the strainhardeningeffect due to the intenseplastic deformation of the particles upon impact with the substrateplayed an important role on increasing the aluminum hardness asreported in other studies [23].

3.3. Heat treatments of the coatings

3.3.1. Intermetallic phaseAl–Mg binary phase diagram is shown in Fig. 6 [31]. It can be seen

that all the Al–Mg intermetallic compound have a melting temper-ature lower than 450 °C and the Mg17Al12 phase has a widercomposition range than Al3Mg2 phase at elevated temperature. Thecold spray coating microstructure after the 4 h heat treatment at400 °C is shown in Fig. 7a. Four distinctive zones can be observed. Thefirst (the top layer) is the un-reacted as sprayed aluminum coating.The second and the third zones represent the intermetallic layers thatformed at the coating/substrate interface. The fourth is the AZ91Dmagnesium substrate shown at the bottom. According to the Al–Mgbinary phase diagram (Fig. 6), it is expected that the intermetalliccompound near the substrate is Mg rich phase -Mg17Al12 (β phase)while the intermetallic layer present near the un-reacted Al coatingwill be the Al rich phase- Al3Mg2 (γ phase).

Quantitative phase's composition was investigated using the EDStechnique. It is accepted that in Al–Mg alloy systems, the compositionranges between 52–60 at.% Mg consist of a single β phase, while γ andR phases will be formed in the area range between 38.5–40.3 and41.7–42.3 at.% Mg respectively. Note that the later, R phase can onlyexist between 320–370 °C, as it decomposes to β and γ phases [31].

Fig. 5. OM image of the etched as-sprayed coating showing the morphology of the Alparticles and the arrows indicated the particles apparently larger than 50 μm.

Three EDS analysis points were performed on the four different layers.According to the composition results, the intermetallic compoundsnear the magnesium substrate and the un-reacted aluminum coatingwere β phase (51.97–55.41 at.% Mg) and γ phase (39.23–39.27 at.%Mg), respectively. Moreover, EDS elemental line scan from themagnesium substrate side to the un-reacted aluminum coating sidewas conducted and the obtained quantitative composition is shown inFig. 7b. It is observed that the β phase has a larger composition rangecompared to the γ phase and that the composition changes abruptly

Fig. 7. (a)SEM image of cold sprayed pure Al onto AZ91D- T4 substrate after heattreatment under vacuum at 400 °C- 4 h.revealing different intermetallic phases and(b) Quantitative EDS line scan as indicated by arrow in (a).

Fig. 9. Growth curve of the intermetallic β- and γ-phases at 400 °C vs. square rootdiffusion time (h1/2).

4669H. Bu et al. / Surface & Coatings Technology 205 (2011) 4665–4671

at the interface of the two layers. It is worth noting that some porositywithin the top layer of the γ phase was observed (arrows in Fig. 7a).Compared to Spencer [15] who also carried out heat treatment on theAl/Mg system, the volume of porosities on the coating cross section inthe current study is smaller and only a few pores are observed.

3.3.2. Intermetallic thicknessThe intermetallic thickness variation with heat treatment time is

shown in Fig. 8. It can be observed that the γ phase and β phase growsteadily for heat treatment at 400 °C under 16 h. For the 20 h heattreatment, the thickness of the γ phase layer increases at a muchslower rate while the β phase does not show noticeable growth. Theseresults are similar to previous studies [20], where a post annealing onAl/Mg at 413 °C was conducted, and found that the two intermetalliclayers did not grow noticeably when the heat treatment timeexceeded 24 h. Comparable conclusions can also be drawn from thelinear regression growth curve shown in Fig. 9, where it can beobserved that the slope of the γ phase is larger than β phase, and theformer is almost 2.5 times higher than the latter. The curves reveal alinear trend that goes through the origin point, which demonstratesthat the diffusion process obeys a parabolic rate law for heattreatment time no more than 20 h. Similar results have also beenreported by Y. Funamizu and K. Watanabe [32] who suggested thatthe growth rate of the γ phase and β phase represent a linear relationwith the square root of diffusion time when the heat treatment wasperformed at 425 °C and the slope of γ phase is almost 2.2 times ofthat for β phase.

It implies that the intermetallics grow faster at the initial stage andthen the overall concentration gradient decreased swiftly associatedwith the intermetallic thickness. It can be found in Fig. 8 that themigrating distance of Al atoms into Mg substrate is a little more than60 μm, compared to less than 60 μm for the Mg atoms diffusing in thereverse directionwhen the heat treatment time is 4 h. This suggests thatthe aluminum diffuses more rapidly than magnesium in the system.Moreover, it is noted thatwhen increasing the heat treatment time from4 h to 20 h, the position of the β/γ interface is almost stable, whichimplies that these two intermetallics grow individually other thantransform fromone to the other. The intermetallic growth rate ismainlydue to the atom diffusion rate in the two intermetallics. Thus, one of thereasons for higher growth rate of γ phase during the heat treatment isprobably due to the higher diffusion rate of aluminum atoms in the twointermetallic phases [20,32]. Note that when two or more intermediatephases are formed in a diffusion zone in a binary system suchAluminum–Magnesium, many factors control the occurrence andgrowth of the various phases. The growth of a phase, in multiphase

Fig. 8. Thickness measurement of the intermetallic layers vs holding time of the heattreatment at 400 °C under vacuum. The horizontal zero axis corresponds to the originalMg/Al interface before the heat treatment.

diffusional growth is dependent on its own interdiffusion coefficient aswell as those of its neighboring phases. Funamizu et al. [32] haveattributed the difference in the thickness growth of the β and γ phaseduring heat treatment to the difference between the activationenergy of layer growth and that of the interdiffusion of the two phases.The later was attributed to the difference in the composition rangeand temperature dependence of the two intermetallic phases as shownin the equilibrium phase diagram of the Al–Mg system (Fig. 6).High interdiffusion coefficient and low temperature dependence ofthe γ-phase composition contribute significantly in the growth of thisintermetallic as compared to β-phase. More details about the differentactivation energies of the different phases within the Al–Mg system canbe found elsewhere [32].

According to Fick's Second Law, the atom diffusion coefficient ismainly due to the concentration gradient and the temperature, andthe diffusion distance is proportional to the square root of the heattreatment time. The thickness of the total intermetallics thickness-diffusion distance-in this study obeys a parabolic rate law [32] (Fig. 9).However, Spencer et al. [15] suggested that the parabolic law is notapplicable in the Al/Mg couple diffusion process because of theKirkendall effect, which will lead to the condensation of vacanciesnear the edge of γ phase and then decrease the diffusion rate. It can beseen in Fig. 7a that there are few small pores near the edge of γ phase,as opposed to a large number of holes observed in Ref. [15] at thesame position after diffusion, despite similar porosity level content inthe as-sprayed coatings. The reason for the smaller porosities at the γphase/un-reacted Al coating may be due to the large Al particleswhich were used as feedstock powder resulting in a coating withreduced number of Al/Al splats that would enhance the Mg diffusionin to the γ phase.

As the thickness of the intermetallics obeys the parabolic law,increasing the heat treatment time or temperature to obtain thickerintermetallics is not a solution. First, the higher temperature willcause the melting of the intermetallic phases which were formed bythermal diffusion [4]; moreover, it will accelerate the grain growth ofmagnesium substrate which is detrimental to the mechanicalperformance based on the Hall–Petch relationship [33]. It has beenrecommended that the Al/Mg couple diffusion-induced phase shouldbe performed between 400 °C to 436 °C [15], taking into account thatthe eutectic β phase will start local dissolution at 426 °C [5]. As such,the diffusion temperature was limited in a narrow extent and in thecurrent study a temperature of 400 °C was selected, because it is thesame temperature for AZ91D magnesium alloy solution treating.Consequently, during the diffusion heat treatment, the β phase willnot precipitate and the substrate will remain a single α phase.

Fig. 11. Comparison of the intermetallic thicknesses formed on the coating whichsprayed onto as cast and T4 substrates heat treated at 400 °C during 4 h.

4670 H. Bu et al. / Surface & Coatings Technology 205 (2011) 4665–4671

When the as-sprayed coating samples are heat treated, the atomscan get enough energy and diffuse resulting in the formation of Al–Mgbinary alloy band [20] between the Al coating and the Mg substrate.Generally, this binary band should contain αAl, αMg, γ phase and βphase with a concentration gradient according to the diffusion theory[22], but in fact it was found that it mainly consists of two differentintermetallic compounds. Similar to other work [20], the layers of Mgmatrix with saturated Al atoms and Al matrix with saturated Mgatoms were not detected in the OM and SEM images due to thelimitation of the technique used in this work. Because Al/Mg diffusionis a reaction diffusion system, Al atoms and Mg atoms can react andform only one intermetallic compoundwhen the atomic ratio arrive ata critical value at some particular point, according to the second law ofthermodynamics. That is why the composition between the in-termetallics and substrate/β phase and unreacted Al/γ phase will bechanged abruptly rather than gradually.

The typical optical microstructure of the as sprayed coating afterheat treatment of 20 h is shown in Fig. 10. With a heat treatment timeof 20 h at 400 °C, the β phase was 83 μm thick while the γ phase was149 μm thick for a total thickness of 232 μm. Compared to otherstudies [20] which attained a total thickness of a little more than160 μm below 413 °C for 24 h, the total intermetallic thickness isincreased by 45%.

The main reason for the thicker intermetallic layers found in thecurrent work is believed to be that T4 instead of as-cast magnesiumbulk [15] was used as substrates. To confirm this assumption, acomparison of the intermetallic thickness on T4 and as-cast AZ91Dafter 400 °C heat treatment for 4 h is shown in Fig. 11. The β phase andγ phase thicknesses were respectively 36.7 μm and 93.3 μm for T4substrate and 30.8 μm and 83.2 μm for as-cast substrate. Generally,the aluminum content in the AZ91 magnesium bulk is near 9 wt.%. T4substrates have a single solid solutionα-phase with very fine clusters/GP zones and Al atoms concentration is higher and more homoge-neously distributed within Mg matrix. Hence intermetallic growth isfaster as less enrichment is required to reach the critical level forintermetallic formation in the substrate. However, as-cast AZ91substrates contain β phase precipitates and α phase Mg matrix. Theβ phase precipitates have aluminum content of 44.3%, leading to amagnesiummatrix containing less than 9% aluminum because of betaprecipitation. Thus, the as-cast state has more Al tied up in the betaphase and consequently less available in the Mg matrix to formintermetallics. Moreover, it is expected that the dense coating and the

Fig. 10. OM image of the cold sprayed Al coating onto AZ91D-T4 substrate after heattreatment at 400 °C for 20 h.

intimate contact of the Al/Mg interface and the Al splats, which weredramatically influenced by the feedstock powders and the sprayingconditions, were believed to have a positive effect on Al diffuseprocess.

3.3.3. HardnessThe purpose of cold spray coatings and the intermetallics formed

by post heat treatment is providing a protection of magnesiumsubstrates from the environment. Any notable degradation of thesubstrate due to the heat treatment should be avoided. As such,microhardness testing was used to evaluate the mechanical proper-ties of the Mg substrates. The microhardness of the Mg substrate andof the two intermetallic layers as a function of heat treatment time isshown in Fig. 12. The hardness value of as cast and T4 Mg substrateswere 80.7±1.8 HV0.025 and 73.7±4.0 HV0.025, respectively. Thishardness decrease was mainly due to the dissolution of the hard βphase, which acts as a reinforcement phase in the AZmagnesium alloyseries, in the Mg matrix. After the diffusion heat treatment, the βphase was precipitated from the Mg matrix which saturated with Alatoms during the cooling process, and the hardness of Mg substrateincreased to around 91 HV0.025. Despite the diffusion treatment beingcarried out for different holding time, the Mg substrate hardness wasalmost the same, because the substrate maintains a single α phasewhen the furnace temperature is kept at 400 °C and the β phase couldnot precipitate in this process. All the samples in this study were

Fig. 12. Microhardness results of the Mg substrates and the intermetallic layers.measured after 400 °C heat treatment for different period of time.

4671H. Bu et al. / Surface & Coatings Technology 205 (2011) 4665–4671

cooled from 400 °C to room temperature according to the samecooling condition. Therefore, it is expected that the Mg substratesshould have the same β phase content which is considered as areinforcement phase in AZ series Mg alloys and could only precipitatein the cooling process. Moreover, the hardness of the γ phase wasslightly larger than the β phase, 279.6±13.7 HV0.025 and 260.5±10.7HV0.025 respectively. Similar values have been reported by Spencer etal. [15]. Note that the measured hardness is 2 times higher than the ascast AZ91Dmagnesium substrate. It is widely accepted that the higherhardness of the coating is considered helpful to attain a higher wearresistance [34].

4. Conclusion

The cold spray parameters and nozzle used in this study areappropriate for spraying large aluminum powders with wide sizedistribution and a dense coating is attained which has a porositycontent less than 1%. The growth rate of the Al–Mg intermetallicphases during annealing at 400 °C obeys a parabolic law and the Al-enrich phase (γ phase) grows much faster- around 2.5times- than theMg-enrich phase (β phase). Increased diffusion time will lead tothicker intermetallics between un-reacted Al coating and Mgsubstrate. The total thickness of the intermetallics after 20 h diffusionis more than 230 μm, which is 45% thicker than thicknesses that havebeen reported in the literature. Considering the relative lowertemperature and holding time used in this study, it is believed thatthe coating can provide better protection and less detrimental effectfor the Mg substrate.

Compared to the AZ91Dmagnesium alloy bulk, nomatter as cast orT4 status, the intermetallic compounds which formed after cold sprayand post heat treatment have a much higher hardness, increasing thepossibility of promoting the wear resistance of magnesium alloysurface. As a result, the pure Al coatings deposited on Mg substratesusing cold spray and post heat treatment can be treated as an alter-native method for magnesium protection.

Acknowledgments

The authors would like to thank the Meridian Technologies(Ontario, Canada) for providing the AZ91D Magnesium materialsand Shanghai JiaoTong University for the financial support given tothis work.

References

[1] B.L. Mordike, T. Ebert, Mater. Sci. Eng. A 302 (1) (2001) 37.[2] B. Deforce, T.J. Eden, H.W. Pickering, Mater. Perf. 48 (2) (2009) 40.[3] A. Pardo, M. Mohedano, P. Casajus, A.E. Coy, R. Arrabal, Surf. Coat. Technol. 203 (9)

(2009) 1252.[4] I. Shigematsu, M. Nakamura, N. Saitou, K. Shimojima, J. Mater. Sci. Lett. 19 (6)

(2000) 473.[5] T. Zhu, W. Gao, Mater. Sci. Eng. 4 (2009) 1.[6] V.K. Champagne, J Fail. Anal. Preven. 8 (2) (2008) 164.[7] Q.Wang, K. Spencer, N. Birbilis, M.X. Zhang, Surf. Coat. Technol. 205 (1) (2010) 50.[8] J.E. Gray, B. Luan, J. Alloy. Comp. 336 (1–2) (2002) 88.[9] A.P. Alkhimov, A.N. Papyrin, V.F. Kosarev, US Patent 5 302 414 (1994).

[10] T. Novoselova, P. Fox, R. Morgan, W. O'Neill, Surf. Coat. Technol. 200 (8) (2006)2775.

[11] M. Grujicic, J.R. Saylor, D.E. Beasley, W.S. DeRosset, D. Helfritch, Appl. Surf. Sci. 219(3) (2003) 211.

[12] H. Assadi, F. Gartner, T. Stoltenhoff, H. Kreye, Acta Materialia 51 (15) (2003) 4379.[13] B. Jodoin, J. Therm. Spray Technol. 11 (4) (2002) 496.[14] W.Y. Li, C. Zhang, X.P. Guo, G. Zhang, H.L. Liao, C. Coddet, Appl. Surf. Sci. 253 (17)

(2007) 7124.[15] K. Spencer, M.X. Zhang, Scripta Materialia 61 (1) (2009) 44.[16] A.C. Hall, R.A. Neiser, T.J. Roemer, D.A. Hirschfeld, J. Therm. Spray Technol. 15 (2)

(2006) 233.[17] K. Spencer, D.M. Fabijanic, M.X. Zhang, Surf. Coat. Technol. 204 (3) (2009) 336.[18] H.Y. Lee, S.H. Jung, S.Y. Lee, K.H. Ko, Mater. Sci. Eng. A 433 (1–2) (2006) 139.[19] T. Novoselova, S. Celotto, R. Morgan, P. Fox, W. O'Neill, J. Alloy. Comp. 436 (1–2)

(2007) 69.[20] M.X. Zhang, H. Huang, K. Spencer, Y.N. Shi, Surf. Coat. Technol. 204 (14) (2010)

2118.[21] T. Stoltenhoff, H. Kreye, H.J. Richter, J. Therm. Spray Technol. 11 (4) (2002) 542.[22] P. Liu, Y. Li, H. Geng, J. Wang, H. Ma, G. Guo, Metall. Mater. Trans. B. 37 (4) (2006)

649.[23] T.H. Van Dteenkeste, J.R. Smith, R.E. Teets, Surf. Coat. Technol. 154 (2–3) (2002)

237.[24] T. Schmidt, F. Gartner, H. Assadi, H. Kreye, Acta Materialia 54 (3) (2006) 729.[25] T. Schmidt, F. Gartner, H. Kreye, J. Therm. Spray Technol. 15 (4) (2006) 488.[26] S. Cadney, M. Brochu, P. Richer, B. Jodoin, Surf. Coat. Technol. 202 (12) (2008)

2801.[27] L. Ajdelsztajn, A. Zuniga, B. Jodoin, E.J. Lavernia, Surf. Coat. Technol. 201 (6) (2006)

2109.[28] G. Song, A.L. Bowles, D.H. StJohn, Mater. Sci. Eng. A 366 (1) (2004) 74.[29] L. Ajdelsztajn, B. Jodoin, G.E. Kim, J.M. Schoenung, Metall Mater Trans A 36A (3)

(2005) 657.[30] R. Morgan, P. Fox, J. Pattison, C. Sutcliffe, W. O'Neill, Mater. Lett. 58 (7–8) (2004)

1317.[31] J.L. Murray, Bulletin Alloy Phase Diagrams 3 (1) (1982) 60.[32] Y. Funamizu, K. Watanabe, Trans. JIM 13 (1972) 278.[33] A. Yamashita, A. Horita, T,.G,. Langdon, Mater. Sci. Eng. A300 (2001) 142.[34] J.C. Lee, H.J. Kang, W.S. Chu, S.H. Ahn, Annals CIRP 56 (1) (2007) 577.