Pt

Aa

b

c

a

ARRA

KPTMMM

1

cthhirtoapLrawneds

t

0d

Journal of Materials Processing Technology 209 (2009) 5060–5067

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

journa l homepage: www.e lsev ier .com/ locate / jmatprotec

ulsed laser surface treatment of magnesium alloy: Correlation betweenhermal model and experimental observations

noop N. Samanta, Baoshuai Dua,b, Sameer R. Paital a, S. Kumarc, Narendra B. Dahotrea,∗

Department of Material Science and Engineering, The University of Tennessee, Knoxville, TN 37996, USASchool of Materials Science and Engineering, Shandong University, Jinan 250061, ChinaDepartment of Materials Engineering, Indian Institute of Science, Bangalore 560012, India

r t i c l e i n f o

rticle history:eceived 31 October 2008eceived in revised form 20 January 2009ccepted 7 February 2009

eywords:

a b s t r a c t

The effect of deposition of Al + Al2O3 on MRI 153 M Mg alloy processed using a pulsed Nd:YAG laser ispresented in this study. A composite coating with metallurgical joint to the substrate was formed. Themicrostructure and phase constituents were characterized and correlated with the thermal predictions.The laser scan speed had an effect on the average melt depth and the amount of retained and/or reconsti-tuted alumina in the final coating. The coating consisted of alumina particles and highly refined dendritesformed due to the extremely high cooling rates (of the order of 108 K/s). The microhardness of the coating

ulsed laserhermal modelingRI 153 M Mg alloyicrohardness

was higher and several fold improvement of wear resistance compared to the substrate was observedfor the coatings. These microstructural features and physical properties were correlated with the effectspredicted by a thermal model.

elt depth

. Introduction

The increasing demand of light structural materials in commer-ial applications, such as automobile and aerospace, has drivenhe rapid development of Mg alloys due to their low density andigh specific strength (Cao et al., 2006). However, the inherentexagonal close-packed (HCP) crystal structure of Mg and its inabil-

ty to form a dense oxidation layer on the surface lead to theestricted mechanical properties of Mg alloys and thereby limitheir applications in wear and corrosion environments. In lightf these limitations, modification of the bulk materials throughlloying and forming Mg-matrix composites (fiber reinforced andarticulate reinforced) has been attempted (Volkova, 2006; Ye andiu, 2004). Nevertheless, the improvement of wear and corrosionesistance of Mg alloys can also be achieved by modifying chemicalnd/or physical properties of the surface, since the failure throughear and corrosion often begin at the surface of components. Aumber of coating technologies (e.g. CVD, PVD, plasma spraying,lectron beam vaporization, magnetron sputtering, microwave oxi-

ation and laser assisted technique) have been tried to modify theurface properties of Mg alloys (Gray and Luan, 2002).

Among various surface modification techniques, laser surfacereatment of Mg alloys has attracted significant interests. The key

∗ Corresponding author. Tel.: +1 865 974 3609; fax: +1 865 974 4115.E-mail address: ndahotre@utk.edu (N.B. Dahotre).

924-0136/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.jmatprotec.2009.02.004

© 2009 Elsevier B.V. All rights reserved.

feature of this process is localized melting and solidification withina short time and shallow depth, while the bulk material remainscool and serves as an infinite heat sink. Thus, a wide variety of chem-ical and microstructural states can be retained owing to this rapidquench effect (Draper and Ewing, 1984). Moreover, short process-ing time, flexibility in operation, economy in time/energy/materialconsumption, shallow heat affected zone and precision are theimportant advantages of laser surface engineering over conven-tional processes (Majumdar et al., 2004). In the past, laser remeltingand cladding has been reported to improve the wear and corrosionresistance of Mg and its alloys (Abbas et al., 2005; Mondal et al.,2008; Yue et al., 2007; Yue and Su, 2007; Yue and Lin, 2008; Wanget al., 1993). It is also found that hard ceramic reinforcement can beintroduced into metal matrix to form a composite coating on Mgalloys by simultaneous laser melting of a precursor and a portionof the substrate. Composite coatings reinforced by Al2O3, SiC andMg2Si have been deposited on Mg alloys by laser surface engineer-ing and it was shown that by forming such a composite coating wearand corrosion resistance can be significantly improved (Majumdaret al., 2004, 2003; Gao et al., 2006; Volovitch et al., 2008). Althoughcontinuous wave laser has been extensively investigated in terms oflaser surface modification of Mg alloys, there is a lack of study aboutpulsed laser. Compared with continuous wave laser, the high peak

power of a single pulse produced by the pulse laser can readily reachthe threshold of high absorbtivity of Mg which is around the meltingpoint, thus resulting in better absorption. The short duration timeof laser pulse which means short heating time can also contribute

Proces

tttcp

sicot(tsmwrc

2

ZspspbT1dabloaailAo80c

P(actcwcfaet1N

tsthmc

A.N. Samant et al. / Journal of Materials

o higher cooling rates and subsequent grain refinement and forma-ion of non-equilibrium phases. Besides, by adjusting the durationime, repetition rate, pulse energy, a flexible processing windowan be obtained for tailoring and patterning the microstructure androperties of the coating.

The aim of this study was to examine the effect of processingpeed on the depth of coating and explore the feasibility of improv-ng wear resistance of Mg alloy using a pulsed laser by forming aeramic particle reinforced composite coating. A powder mixturef Al2O3 (95 wt.%) + Al (5 wt.%) was used as the precursor owing tohe ultra hardness and superior tribological properties of aluminaDong and Bell, 1999). 5 wt.% of aluminum was added to improvehe interfacial strength and the wetting behavior between magne-ium and alumina (Majumdar et al., 2004). Phase constituents andicrostructure was characterized. Furthermore, thermal modelingas carried out to understand the effects of thermal cycle expe-

ienced by the materials during the laser treatment and furtherontrol over such processing.

. Experimental

Permanent mould cast (PM) MRI 153 M alloy (Al 7.95, Ca 0.98,n < 0.01, Mn 0.2, Sr 0.27, Sn <0.01, Mg bal. wt.%) plates with dimen-ions of 45 mm × 45 mm × 4 mm were used as a substrate. Therecursor comprising of 95 wt.% Al2O3 (99.0% purity, polygonal inhape with an average particle size of 10 �m) and 5 wt.% Al (99.5%urity, 10 �m particle size) was mixed with a proprietary organicinder (LISI WI5853, Warren Paint and Color Company, Nashville,N) and spray deposited to an average thickness of approximately20 �m on the substrate. The Al2O3 and Al particles were evenlyistributed in the precursor. JK 701 pulsed Nd:YAG laser with anverage power of 400 W was used for laser processing. The laseream was delivered by a fiber optic system and a 120 mm focal

ength convex lens providing a defocused spot diameter of 500 �mn the surface of substrate. The configuration of the focusing lensssembly provided a uniform beam distribution, in both temporals well as spatial evolution. Hence it was assumed that the energys evenly distributed across the pulse. Several laser tracks wereaid with minimal overlap (<5%) to cover the entire surface area.s the overlap was very small, no significant overlap effects werebserved. The scanning speeds of 21 mm/s, 42 mm/s, 63 mm/s, and4 mm/s were employed for laser processing at a pulse width of.5 ms, repetition rate of 20 Hz, and pulse energy of 4J using Ar as aover gas at a pressure of 80 psi (5.5 bar).

Structural characterization was conducted by employing ahilips Norelco X-ray Diffractometer (XRD) with CuK� radiationwavelength 1.5418 Å) operating at 20 kV and 10 mA. The scanningngle (2�) ranged from 20◦ to 100◦ with a step size of 0.02◦ andount time of 1 s. Samples for microstructure observation were cutransverse to the laser track by a low speed diamond saw under oil-ooling condition. These samples were then successively polishedith 600, 800, and 1200 grit emery papers. Final polishing was

onducted with 0.3 �m alumina suspension on a microcloth andollowed by etching with a mixture of 100 ml ethanol, 10 ml aceticcid, 6 ml picric acid, and 20 ml distilled water. A LEO 1525 scanninglectron microscope (SEM) was used to observe the microstruc-ure. Microhardness tests were performed with a load of 50 g for2 s duration by a Vickers microhardness tester (Buehler, Modelumber: 1600-6300).

Wear tests were performed by using a pin-on-disc wear and fric-ion monitor (Model: No. TR201EV) according to the ASTM G-99

tandard. The experiment was conducted under dry sliding condi-ion at room temperature. The disc with diameter of 100 mm andardness of HRC 58 was made from EN 32 steel. Cylindrical speci-ens of 6 mm diameter and 5 mm height were machined from the

oated sample and used as the pin. During the wear testing, the pin

sing Technology 209 (2009) 5060–5067 5061

was placed at a distance 40 mm from the center of the disc rotatingat 200 rpm, corresponding to a linear velocity of 0.837 m/s. In thepresent investigation the total sliding distance was kept constantat 1 km while the load was varied from 5 N to 40 N. The wear rate(mm3/m) was presented as the total volume lost within the weartrack over the entire sliding distance.

3. Computational

The various physical attributes such as depth of melting andother microstructural features during laser processing depend onthe temperature distribution and associated cooling rates throughthe coating and substrate. For this purpose, a thermal model wasbuilt in COMSOL’sTM heat transfer transient mode to correlate theobserved attributes with the laser processing conditions (Harimkaret al., 2006; Harimkar et al., 2007; Basu et al., 2008; Du et al., 2008a;Chen et al., 2008). Geometry in the form of a cube correspond-ing to the precursor (45 mm × 45 mm × 120 �m) was coupled withanother cube representing the substrate (45 mm × 45 mm × 4 mm).The temperature evolution and cooling rates are governed by theenergy input to the system and the time for which the surface isexposed to this laser energy (irradiation time). The repetition rateand the pulse width of the laser beam of diameter, 500 �m weretaken into account to determine the irradiation time because therewas a power off period between two successive pulses. The resi-dence time was estimated using the following relation (Samant etal., 2007):

Residence time (s)

= Repetition rate (Hz) × Pulse width (s) × Beam diameter (cm)Traverse speed (cm/s)

(1)

Laser processing speeds of 21 mm/s, 42 mm/s, 63 mm/s, and84 mm/s corresponded to irradiation times of 236.3 �s, 118.0 �s,78.8 �s, and 59.1 �s respectively. Thus, in this study, the laser beamwas considered a quasi-stationary heat source and the coordinatesystem could be translated along with this heat source. The effectof a moving laser beam was considered to be the same as a sta-tionary beam that is interacting with the surface for a time equalto the residence time. Similar approach has been considered in thepast by Harimkar et al. (2006), Du et al. (2008a), Chen et al. (2008),Basu et al. (2008), Ahlström et al. (2004), Arnold et al. (1993) andPatwa and Shin (2007). The residence time predicted for differentspeeds was input to the model along with the energy density perunit beam cross-sectional area given by:

Energy density (W/cm2)

= Incident energy (J)Pulse width (s) × Area of incident beam (cm2)

(2)

Assuming a uniform beam distribution in both the temporal andspatial evolution, the energy density corresponding to input energyof 4 J, pulse width of 0.5 ms and beam cross-sectional area of1.96 × 10−3 cm2 was 4.07 × 106 W/cm2.

The heat transfer taking place during laser processing was mod-eled using Fourier’s second law of heat transfer:

∂T(x, y, z, t)∂t

k(T)[

∂2T(x, y, z, t) ∂2T(x, y, z, t) ∂2T(x, y, z, t)]

=�Cp(T) ∂x2

+∂y2

+∂z2

(3)

where k(T) and Cp(T) are the variations in thermal conductivity andspecific heat as a function of temperature (Fig. 1), � is the den-sity, T is the temperature field, t is the time and x, y and z are the

5062 A.N. Samant et al. / Journal of Materials Proces

Fsp

sctg(mpaabschpo

a

wi(taotttatftrapIoiacw

ig. 1. Variation of thermal conductivity and specific heat of the precursor andubstrate with temperature. Inset: Variation of heat transfer coefficient with tem-erature.

patial directions. Using the law of mixtures, the density of pre-ursor was computed to be 3278 kg/m3 that also took into accounthe porosity of the precursor layer (∼12% from the optical micro-raphs using Image JTM software) while the density of the substrateMRI 153 M Mg) was found to be 1774.24 kg/m3. As laser surface

elting is a rapid heating and cooling process, the thermophysicalroperties of materials under laser irradiation change rapidly overlarge temperature range. Hence, variation of thermal conductivitynd specific heat were considered as a function of temperature foroth the precursor and the substrate for improving the accuracy ofimulations (Fig. 1) (Touloukian, 1967). The latent heat of solidifi-ation was accounted for by incorporating the variation of specificeat as a function of temperature. This took into consideration thehase change due to melting. At time t = 0, the initial temperaturef T = T0 = 300 K was applied.

The balance between the absorbed laser energy at the surfacend the losses due to radiation was given by:

−k(T)

(∂T(x, y, 0, t)

∂x+ ∂T(x, y, 0, t)

∂y+ ∂T(x, y, 0, t)

∂z

)

= ıaI − ε�(T(x, y, 0, t)4 − T40 )

ı = 1 when 0 ≤ t ≤ tr

ı = 0 when t > tr

(4)

here I is the laser energy density predicted above in Eq. (2), k(T)s the temperature dependent thermal conductivity of the materialFig. 1), ε is the emissivity for thermal radiation, tr is the irradiationime (Eq. (1)), � is Stefan–Boltzman constant (5.67 × 10−8 W/m2K4),nd a is the absorptivity of the material. The term ı takes a valuef 1 when time, t is less than irradiation time, tr and it is 0 whenime, t exceeds irradiation time. Thus the value of ı depends onhe pulse on time and ensures that the energy is input to the sys-em only during the irradiation time and cuts off the energy supplyfter that during pulse off time. The laser energy is incident onhe precursor and it is then transferred to the substrate by dif-erent heat transfer phenomena. Hence, only the absorptivity ofhe precursor was considered for calculations. Furthermore, as theeflection of a mixture of opaque particles is a function of the rel-tive surface area (Schatz, 1967), the surface-area percent of therecursor was used to evaluate the absorptivity of the precursor.

n order to determine the surface-area percent, the weight percent

f the components of the precursor (Al2O3 and Al) was convertednto volume percent. The size of all the particles being same, it wasssumed that volume percent was the same as surface-area per-ent. Thus, considering 12% porosity, 95 wt.% Al2O3 and 5 wt.% Alere equivalent to 82 and 6 area % respectively. In situ absorptiv-

sing Technology 209 (2009) 5060–5067

ity measurements in short duration high energy density dynamicprocess such as laser–material interaction being extremely diffi-cult, values of absorptivity of precursor components (0.8 for Al2O3and 0.17 for Al at 1.06 �m wavelength) commonly found in lit-erature (Touloukian, 1967) along with their surface-area percentmentioned above were used for calculations yielding an absorp-tivity value for the precursor of 0.66. Nonetheless, attempts areongoing for measuring absorptivity under laser processing con-ditions similar to the ones employed in this study and they willbe incorporated in future simulations. Also, as for a given materialand processing condition, the absorptivity is equal to the emissivity(Sturge, 2003), emissivity was also taken as 0.66.

The convection at the bottom surface of the sample was givenby:

−k(T)

(∂T(x, y, D, t)

∂x+ ∂T(x, y, D, t)

∂y+ ∂T(x, y, D, t)

∂z

)

= h(T(x, y, L, t) − T0), (5)

where D is the thickness of the sample which was taken as 4.12 mm(120 �m for the precursor +4 mm for the substrate) and h(T) is theheat transfer coefficient (W/m2 K) which was included as a func-tion of temperature (inset in Fig. 1) (Incropera and Dewitt, 2002).Even though the sample was sufficiently thick (4.12 mm), it was notassumed to be a semi-infinite body and convection from bottomsurface was not neglected in order to improve accuracy of calcula-tions by considering all modes of heat transfer on the laser processas also done by Shuja and Yilbas (2000) and by Yilbas and Mansoor(2006). A non-uniform grid was used for simulations with a finermesh under the laser beam as compared to the rest of the geome-try where temperatures and their gradients are highest (Patwa andShin, 2007). The grid independence test was conducted and meshsize of approximately 53,000 elements resulted in a grid indepen-dent solution. The temperature profiles were obtained by runningthe simulations with extremely small time steps of 10 �s in orderto incorporate the changes in surface temperature during the resi-dence time due to laser scanning speed. The corresponding coolingrates (slope of the cooling curve at different time instants) anddepth of melting were obtained from these computations and theircorrelations with the processing conditions are discussed in thelater part of this study.

4. Results and discussion

The microstructure and structural characterization of the laserprocessed samples was carried out and an attempt was made tocorrelate them with the thermal modeling predictions as discussedin this section.

4.1. Structural characterization

Structural characterization of the coating was carried out inorder to see the effect of laser processing on the phase distributionand corresponding microstructure in the surface and subsurfaceregions. In light of this, the overlay of XRD patterns obtained fromthe surface of the sample is shown in Fig. 2. The major phasespresent in the coating are �-Mg and Al2O3, and some minor peaksbelonging to Al0.9Mg3.1 and MgO were also detected. In general,the coatings produced using different laser scan speeds have simi-lar XRD patterns, and only the peak intensity of each phase varied,illustrating that a relatively broad processing window can be used

without affecting the phase constituents of the composite coat-ing considered in the present work. The laser–material interactionleads to the melting of the precursor (Al + Al2O3) and a portion ofthe substrate (MRI 153 M Mg). The subsequent cooling and solidifi-cation result in evolution of the phases listed above. The formation

A.N. Samant et al. / Journal of Materials Processing Technology 209 (2009) 5060–5067 5063

o

FtepAmo

Fs

Fig. 2. XRD patterns of the laser processed samples.

f MgO is attributed to the following reaction (Ragone, 1995):

Mg + O = MgO �G = −603, 960 + 142.05T − 5.36T ln T (298 K ≤ T ≤ 923 K)

�G = −608, 140 + 112.76T − 0.44T ln T (923 K ≤ T ≤ 1380 K)(6)

rom the Gibbs free energy of the above reaction, it can be foundhat Mg possesses a high affinity to O, which can come from the

nvironment and/or dissociated/melted Al2O3 due to the high tem-erature attained during the process. However, the presence ofl2O3 in XRD patterns corresponding to all four processing speedsay be due to resolidification of melted Al2O3 and/or reconstitution

f decomposed/dissociated Al2O3 through the following reaction

ig. 4. SEM images of the laser processed samples: (a) 21 mm/s, (b) 42 mm/s, (c) 63 mmecondary dendrites.)

Fig. 3. Relative phase amount as a function of laser processing speed.

(Gaskell, 1995):

2Al(l) + 1.5O2(g) = Al2O3(s)

�G = −1, 687, 200 + 326.8T (993 ≤ T ≤ 2327 K) (7)

A semiquantitative analysis of the relative amount of each phasewas conducted using the following formula (Greish and Brown,2003; Du et al., 2008b):

%Ii = Ii∑41Ii

(8)

where Ii is integrated peak intensity of the phase in concern and∑41Ii corresponds to the sum of integrated peak intensity of each of

the four phases in the present composite coating. The correspond-ing plans of the peaks used in the calculation are indicated in Fig. 2and the result of semiquantitative analysis is shown in Fig. 3. A

/s, and (d) 84 mm/s. (Inset: High magnification image of the coating showing the

5064 A.N. Samant et al. / Journal of Materials Processing Technology 209 (2009) 5060–5067

Table 1Variation of measured melt depth, residence time, diffusion length, maximum cooling rate and microhardness (in coating and substrate) with processing speed.

Speed (mm/s) Average measured meltdepth (�m)

Residence time(�s)

Diffusion length(�m)

Maximum coolingrate × 108 (K/s)

Average microhardnessin coating (HV)

Average microhardnessin substrate (HV)

468

dattsial1patpi

4

psiaal

ttws25mamt

wgntccp1S�p(p(bdfttd

Although, the maximum temperatures reached under all fourprocessing speeds were sufficiently higher than melting point ofalumina (2323 K (Touloukian, 1967)) (Fig. 6), the residence timevaried for all speeds which in turn decides if alumina particles

21 363 ± 100.26 236.3 28.22 317 ± 114.35 118.0 19.93 279 ± 55.79 78.8 16.24 224 ± 68.88 59.1 14.1

ecreasing trend of the Mg-containing phases (�-Mg, Al0.9Mg3.1nd MgO) with the increasing laser scan speed is observed whilehe relative amount of Al2O3 showed a reverse trend. With the facthat the precursor thickness was same (120 �m) for all processingpeeds, the laser material interaction time decreased with increas-ng speed leading to observed trend in the phase content. It alsoppeared that although Mg may have suffered some evaporativeosses during laser processing due to its low boiling point (Li et al.,993), it had a marginal effect on the trend of relative amount ofhases in the coating. As seen later, based on the thermal modelingnd experimental observation, it is found that the measured coatinghickness decreased from 363 �m to 224 �m with increased laserrocessing speed from 21 mm/s to 84 mm/s due to the correspond-

ng decrease in laser–material interaction time.

.2. Microstructural evolution and thermal predictions

Cross-sectional views of the samples treated at different laserrocessing speeds are shown in Fig. 4. The composite coating on theurface has a wavy surface texture due to the overlap pulse process-ng. The distinct interface between the coating and substrate showssound metallurgical joint. Moreover, it can be found that the heatffected zone (HAZ) is practically negligible, which is indicative ofocalized heating associated with the laser beam.

Due to the uneven profile, the depth of melting was measured aten different locations in three different cross-sectional areas andhe average value was reported for each processing speed alongith the standard deviation corresponding to the scatter in mea-

urements (Table 1). An increase in the processing speed from1 mm/s to 84 mm/s reduced the residence time from 236.3 �s to9.1 �s respectively. As the effective laser energy going into theaterial system reduced with residence time, less material melted

nd average melt depth decreased with increased speed. Theseeasured melt depths were compared with values predicted using

emperature distribution profiles as seen later in this study.The coating mainly consisted of polygonal particles (indicated by

hite circle in Fig. 4d) in addition to highly refined cellular dendriticrain structure (insets in Fig. 4). The EDS analysis on these polygo-al particles provided an Al to O ratio close to 0.67, thus confirminghat they are Al2O3 particles from the precursor. The highly refinedellular dendritic grain structure is attributed to extremely highooling rates associated with laser processing. The average com-osition of the cellular dendrite measured by EDS is 83 at.% Mg,1.1 at.% Al, 2.4 at.% O, 1.0 at.% Ca, 1.4 at.% Mn, 0.6 at.% Zn, 0.4 at.%n and 0.1 at.% Sr, illustrating that it is mainly a solid solution of-Mg with other elements. In order to see the effect of the laserrocessing speed on the formation of dendrites, the average SDASsecondary dendrite arm spacing) values corresponding to eachrocessing speed were obtained using the linear intercept methodFig. 5a). This method involves measuring the distances (spacings)etween secondary dendrite arms along a line normal to the den-

rite arms. This procedure was repeated for five different imagerames and the average SDAS is represented in Fig. 5b along withhe corresponding error bars (Harimkar et al., 2006). The correla-ion between the measured SDAS and the predicted cooling rates isiscussed later.

0.96 117.52 ± 3.97 80.87 ± 5.171.23 114.24 ± 8.37 93.1 ± 4.811.89 110.04 ± 12.23 83.56 ± 0.641.98 120.46 ± 9.09 102.4 ± 2.19

The temperature profiles for different processing speedsobtained from Eqs. (1)–(5) above are presented in Fig. 6. Thedecrease in residence time and corresponding input laser energywith increase in processing speed reduced the maximum surfacetemperature. As the temperature distribution into the materialevolves as a function of time, maximum melt depth correspondsto the residence time for each of the processing speeds. The maxi-mum temperatures reached under all four processing speeds aresufficiently higher than the melting point of alumina (2323 K(Touloukian, 1967)). Hence, the total melt depth was predictedby tracing the location of melting point of magnesium (923 K(Touloukian, 1967)), the lowest melting component of the sys-tem) in the temperature distribution within the material (Fig. 7a)because the content of magnesium in substrate (∼90 wt.%) wasmore than the other elements.

Fig. 5. (a) Secondary dendrite arm spacing (SDAS) measurement from a high mag-nification SEM image using linear intercept method. (b) Variation of SDAS withprocessing speed.

A.N. Samant et al. / Journal of Materials Proces

opeptdpa

Fs

Fig. 6. Surface temperature profiles for different processing speeds.

f average size 10 �m in the original precursor are likely to com-letely melt. Hence in order to determine if the processing speedsmployed in the present work were capable of melting the aluminaarticles through their core (i.e. reaching melting temperature at

he center) in the precursor, the spatial decay in the temperatureistribution (diffusion length) at the melting point of alumina wasredicted for each case and was compared with the radius of thelumina particles (5 �m). For directional heat flow problems, the

ig. 7. (a) Prediction of total (final) melt depth. (b) Prediction of melt depth corre-ponding to the melting point of alumina for different processing speeds.

sing Technology 209 (2009) 5060–5067 5065

diffusion length, lH can be approximated using the thermal diffu-sivity, given by k(T)/�Cp(T), and the residence time tr as follows(Daniel et al., 2008; Bäuerle, 2000):

lH = 2√

tr (9)

At the melting point of alumina its thermal diffusivity is8.43 × 10−7 m2/s, density is 3800 kg/m3 (Touloukian, 1967), ther-mal conductivity is 6.27 W/mK and specific heat is 1958 J/kg K. It isseen from Table 1 that the predicted diffusion length for each pro-cessing speed is much more than the radius of an alumina particle.Thus, for all the cases considered in this study, temperatures higherthan the melting point of alumina were attained at the core (cen-ter) of the particle and alumina seen in the melted region (Fig. 2) forall speeds; may have reconstituted, as mentioned earlier, throughresolidification and/or reaction between Al and O. Furthermore,majority of such reconstituted alumina particles within the matrixof coating are refined (Fig. 4) due to the rapid cooling rates (asexplained later) associated with all processing speeds employed inthe present work where as few particles are reconstituted as islandsin the near surface region (Fig. 4b and c).

Also, the melting point of alumina which is much higher thanthat of aluminum (933 K (Touloukian, 1967)) was traced to esti-mate the quantity of melted and reconstituted alumina from theoriginal precursor layer under different processing speeds (Fig. 7b).It can be observed from Fig. 7b that for the processing speeds of21, 42, and 63 mm/s the melting temperature of alumina reachedto the depth equal to or greater than 120 �m indicating that allalumina particles within the original precursor layer of 120 �mthickness completely melted and reconstituted in the final meltregion. Where as for the fastest speed (84 mm/s) the melting tem-perature of alumina can be traced only to the depth of 95 �m(Fig. 7b) suggesting that the alumina particles within the bottom25 �m thick precursor layer were retained without undergoingany melting in the final melt region. As the original precursormixture contained 95 wt.% (82 volume%) alumina, only 19.8 wt.%(17 volume%) unmelted alumina was retained in the final meltregion corresponding to the fastest speed (84 mm/s). In essence,due to the extremely rapid cooling rates associated with all laserprocessing speeds (as explained later) first three speeds producedvery fine reconstituted alumina and the fastest speed provided amixture of reconstituted (83 volume%) and retained (unmelted)(17 volume%) alumina (Fig. 4). Thus the laser processing speed can

be varied to control the content of retained and/or reconstitutedalumina in the coating. It can be observed from Fig. 8 that the finalmelt depths predicted using the model proposed earlier were inclose match with experimentally measured values. Furthermore,

Fig. 8. Comparison between experimental and predicted final melt depth.

5 Processing Technology 209 (2009) 5060–5067

amddai

gictwfimmsaui

pcpt

4

ttrmiaduttoM

stmwiw

066 A.N. Samant et al. / Journal of Materials

s explained above, similar to experimental values, predicted finalelt depth also decreased with increase in processing speed. Slight

iscrepancy between predicted and experimental values could beue to non-uniform nature of the melt pool as mentioned earliernd limitations of the software (Image JTM) employed for conduct-ng these measurements.

The slope of the temperature profiles discussed earlier (Fig. 6)ave the cooling rates presented in Table 1. An increase in process-ng speed from 126 cm/min to 508 cm/min increased the maximumooling rate from 0.96 × 108 K/s to 1.98 × 108 K/s. Thus, even thoughhe cooling rates were extremely high (of the order of 108 K/s), thereas a marginal difference in the magnitudes of cooling rates for dif-

erent processing speeds. According to solidification theory, SDASs a function of the cooling rate (Kurz and Fisher, 1998). Hence

arginal variation in cooling rates (Table 1) was responsible forinor changes in SDAS values observed for different processing

peeds (Fig. 5b). Furthermore, the manifestation of marginal vari-tion in these physical attributes of the melted regions producednder the present set of laser processing parameters is also realized

n the mechanical properties of the coating.Thus the size of the melt pool and associated properties of the

recursor could be controlled/modified by varying the laser pro-essing speed and a thermal model can be used to successfullyredict the solidification characteristics during pulsed laser surfacereatment of MRI 153 M Mg alloy.

.3. Microhardness and wear resistance

Microhardness was measured on the cross-sectional planeransverse to the laser track. The variation of microhardness dis-ributions for melt regions corresponding to all processing speedsemained within the same range and, in general the hardness ofelt region was 1.5–2.0 times higher than the substrate (Fig. 9). The

nterface between the melt pool and the substrate (melt depth) islso shown in Fig. 9 for all the processing speeds to illustrate theepth from which there is a drop in hardness corresponding to thenmelted substrate. The average values of the microhardness forhe coating and the substrate are represented in Table 1 along withhe corresponding scatter in data. Thus, the precursor compositionf 95 wt.% Al2O3 and 5 wt.% Al was able to improve the hardness ofRI 153 M Mg alloy.Wear rate at different applied load for both the as-received sub-

trate and laser treated samples are shown in Fig. 10. As expected,

he wear rate increased with the increase in applied load for both

aterials. Compared with the substrate, a substantial reduction ofear rate of the laser treated samples was observed at all loads,

ndicating an improvement of wear resistance for the coatings. Thisas also in accordance with the increase of microhardness in laser

Fig. 9. Microhardness variation for different processing speeds.

Fig. 10. Wear rate of substrate and laser treated samples at different applied load.(Inset: Detailed view of wear rate of the coatings.)

processed coatings (Fig. 9), since hardness plays an important rolein wear behavior for materials and is generally used as a measure-ment of wear resistance (Chen et al., 2009). This improvement ofwear resistance in the laser treated samples could be explainedby considering its different wear mechanisms from that of the sub-strate. It has been shown in the literature (Mondal et al., 2008; Chenet al., 2009) that micro grooving and micro cutting are the dominantwear mechanisms for Mg alloys, and due to the low microhardnessand poor ductility a high wear loss is expected for Mg alloys. Nev-ertheless, for the laser processed coatings which are composite innature, both the hard reinforcement (predominantly Al2O3) and thehighly refined Mg matrix considerably improved the wear resis-tance (Du et al., 2008c). The Al2O3 particles can effectively preventgrooving and ploughing during the wear process, while the highlyrefined dendrites of Mg play the role of strong and ductile metalmatrix.

In order to explore the influence of laser processing parame-ters on the wear behavior of the laser treated samples, a detailedcomparison of the wear resistance of the composite coatings wereshown in the inset in Fig. 10. In comparison, the coatings processedwith laser scan speeds of 21 mm/s, 84 mm/s, and 42 mm/s experi-enced the high, medium and low wear rates respectively. In thecurrent investigation the processing parameter varied was laserscan speed, which is the predominant factor affecting the tem-perature of melting pool and cooling rate. As explained earlier,the level of temperature determined the extent of retained Al2O3particles whereas the cooling rate controlled the level of refine-ment in grain size (SDAS) and the size of reconstituted Al2O3. FromTable 1 it can be seen that the sample processed with 21 mm/s wasassociated with the lowest cooling rate (0.98 × 108 K/s) among allthe laser processed samples and the SDAS of this sample (Fig. 5)showed a relatively higher value. This coating consisted of relativelycoarser dendrites and reconstituted Al2O3 particles, thus experi-encing the high wear rate. Nevertheless, it is interesting to notethat the sample processed with 84 mm/s experienced lower tem-perature (Fig. 4) and the highest cooling rate (Table 1) that wereresponsible for retention of Al2O3 along with formation of refinedand reconstituted Al2O3 leading to medium wear rate. On the otherhand, the samples processed with 42 mm/s (medium cooling rate:0.98 × 108 K/s) contained only refined and reconstituted Al2O3 dueto which it exhibited the lowest wear rate of 0.095 × 10−3 mm3/mat the applied load of 40 N which was approximately 30 times lessthan that of as-received substrate (3 × 10−3 mm3/m). Such contrast

behavior is attributable to the refinement of the grain structure andpresence of retained Al2O3. It is likely that retained Al2O3 particlesmay cause detrimental effect to wear resistance of the compositecoatings. Coating with retained Al2O3 particles may lead to lesswear resistance because of the their coarse size inherited from the

Proces

pmpcirt4aavrbs

5

aaAfftoacTtefiospsat

R

A

A

A

B

BC

C

C

D

Yue, T.M., Su, Y.P., Yang, H.O., 2007. Laser cladding of Zr65Al75Ni10Cu17.5 amorphous

A.N. Samant et al. / Journal of Materials

recursor powder (10 �m) and their poor interface with the Mgatrix compared with that of in situ formed (reconstituted) Al2O3

articles. Such poor interface between particles and matrix mightause pulling out of the particles. As a result, although the coat-ng processed at 84 mm/s is associated with the highest coolingate and finest microstructure, it showed a medium wear resis-ance due to retained Al2O3. On the contrary, coating produced at2 mm/s exhibited the highest wear resistance as it consisted ofcombination of fully reconstituted and refined Al2O3 and it was

ssociated with a moderately high cooling rate. This result also pro-ided an indication that processing parameters that lead to fullyeconstituted Al2O3 particles as well as fine microstructure shoulde considered for tailoring and optimizing the wear properties ofuch composite coating.

. Conclusion

MRI 153 M Mg alloy was successfully coated with 95 wt.% Al2O3nd 5 wt.% Al by using a pulsed Nd:YAG laser. Structural char-cterization showed that the coating consisted of �-Mg, Al2O3,l0.9Mg3.1 and MgO. The average measured melt depth decreased

rom 363 �m to 224 �m as the laser processing speed increasedrom 126 cm/min to 508 cm/min due to the decrease in residenceime (from 236.3 �s to 59.1 �s). Thermal predictions showed thatnly the sample processed with 84 mm/s had retained aluminalong with major portion of reconstituted alumina, while other pro-essing speeds lead to coatings with only the reconstituted alumina.he maximum predicted cooling rate increased from 0.96 × 108 K/so 1.98 × 108 K/s with increase in speed from 21 to 84 mm/s. Thesextremely high cooling rates were responsible for the formation ofne dendrites which had secondary dendrite arm spacing (SDAS)f around 1 �m. Microhardness increased from ∼80 HV for the sub-trate to ∼120 HV for the coating. Due to the presence of hard Al2O3articles and highly refined Mg grains the wear rate decreased sub-tantially compared with that of the substrate (3 × 10−3 mm3/m)nd the sample processed with 42 mm/s showed the best resistanceo wear (0.095 × 10−3 mm3/m).

eferences

bbas, G., Liu, Z., Skeldon, P., 2005. Corrosion behavior of laser-melted magnesiumalloys. Appl. Surf. Sci. 247, 347–353.

hlström, J., Karlsson, B., Niederhauser, S., 2004. Modelling of laser cladding ofmedium carbon steel—a first approach. J. Phys. IV France 210, 405–412.

rnold, N., Kullmer, R., Bäuerle, D., 1993. Simulation of growth in pyrolytic laser-CVDof microstructures-I. One dimensional approach. Microelectron. Eng. 20, 31–41.

asu, A., Samant, A.N., Harimkar, S.P., Majumdar, J.D., Manna, I., Dahotre, N.B., 2008.Laser surface coating of Fe–Cr–Mo–Y–B–C bulk metallic glass composition onAISI 4140 steel. Surf. Coat. Technol. 202, 2623–2631.

äuerle, D., 2000. Laser Processing and Chemistry. Springer, Berlin.ao, X., Jahazi, M., Immarigeon, J.P., Wallace, W., 2006. A review of laser welding

techniques for magnesium alloys. J. Mater. Process. Technol. 171, 188–204.hen, Y., Samant, A., Balani, K., Dahotre, N.B., Agarwal, A., 2008. Effect of laser melting

on plasma sprayed aluminum oxide coatings reinforced with carbon nanotubes.Appl. Phys. A 94, 861–870.

hen, T.J., Ma, Y., Li, B., Li, Y.D., Hao, Y., 2009. Effects of processing parameterson wear behaviors of thixoformed AZ91D magnesium alloys. Mater. Des. 30,235–244.

aniel, C., Armstrong, B.L., Howe, J.Y., Dahotre, N.B., 2008. Controlled evolution ofmorphology and microstructure in laser interference-structured zirconia. J. Am.Ceram. Soc. 91, 2138–2142.

sing Technology 209 (2009) 5060–5067 5067

Dong, H., Bell, T., 1999. Tribological behavior of alumina sliding against Ti6Al4 V inunlubricated contact. Wear 225–229, 874–884.

Draper, C.W., Ewing, C.A., 1984. Laser surface alloying: a bibliography. J. Mater. Sci.19, 3815–3825.

Du, B., Samant, A.N., Paital, S.R., Dahotre, N.B., 2008a. Pulsed laser synthesis ofceramic–metal composite coating on steel. Appl. Surf. Sci. 255, 3188–3194.

Du, B., Paital, S.R., Dahotre, N.B., 2008b. Phase constituents and microstructure oflaser synthesized TiB2–TiC reinforced composite coating on steel. Scr. Mater. 59,1147–1150.

Du, B., Zou, Z., Wang, X., Qu, S., 2008c. Laser cladding of in situ TiB2/Fe compositecoating on steel. Appl. Surf. Sci. 254, 6489–6494.

Gao, Yali, Wang, Cunshan, Lin, Qi, Liu, Hongbin, Yao, Man, 2006. Broad-beam lasercladding of Al-Si alloy coating on AZ91HP magnesium alloy. Surf. Coat. Technol.201, 2701–2706.

Gaskell, D.R., 1995. Introduction to Thermodynamics of Materials. Taylor Francis,Washington, DC.

Gray, J.E., Luan, B., 2002. Protective coatings on magnesium and its alloys—a criticalreview. J. Alloy Compd. 336, 88–113.

Greish, Y., Brown, P., 2003. Phase evolution during the formation of stoichiometrichydroxyapatite at 37.4 ◦C. J. Biomed. Mater. Res. 67, 632–637.

Harimkar, S.P., Samant, A.N., Khangar, A.A., Dahotre, Narendra, B., 2006. Prediction ofsolidification microstructures during laser dressing of alumina-based grindingwheel material. J. Phys. D: Appl. Phys. 39, 1642–1649.

Harimkar, S.P., Samant, A.N., Dahotre, N.B., 2007. Temporally evolved recoil pressuredriven melt infiltration during laser surface modifications of porous aluminaceramic. J. Appl. Phys. 101, 054911.

Incropera, F.P., Dewitt, D.P., 2002. Fundamentals of Heat and Mass Transfer. J. Wiley,New York.

Kurz, W., Fisher, D.J., 1998. Fundamentals of Solidification. Trans Tech Publications,Zurich.

Li, G.H., Gill, H.S., Varin, R.A., 1993. Magnesium silicide intermetallic alloys. Metall.Trans. A 24, 2382–2391.

Majumdar, J.D., Chandra, B.R., Modike, B.L., Galun, R., Manna, I., 2004. Laser sur-face engineering of a magnesium alloy with Al + Al2O3. Surf. Coat. Technol. 179,297–305.

Majumdar, J.D., Chandra, B.R., Galun, R., Modike, B.L., Manna, I., 2003. Laser compositesurfacing of a magnesium alloy with sicicon carbide. Surf. Coat. Technol. 63,771–778.

Mondal, A.K., Kumar, S., Blawert, C., Dahotre, N.B., 2008. Effect of laser surface treat-ment on corrosion and wear resistance of ACM720 Mg alloy. Surf. Coat. Technol.202, 3187–3198.

Patwa, R., Shin, Y.C., 2007. Predictive modeling of laser hardening of AISI5150H steels.Int. J. Mach. Tools Manuf. 47, 307–320.

Ragone, D.V., 1995. Thermodynamics of Materials. J. Wiley & Sons, New York.Samant, A.N., Harimkar, S.P., Dahotre, N.B., 2007. Laser beam operation mode depen-

dent grain morphology of alumina. J. Appl. Phys. 102, 123105.Schatz, E.A., 1967. Reflectance of compacted powder mixtures. J. Opt. Soc. Am. 57,

941–942.Shuja, S.Z., Yilbas, B.S., 2000. 3-Dimensional conjugate laser heating of a moving

slab. Appl. Surf. Sci. 167, 134–148.Sturge, M.D., 2003. Statistical and Thermal Physics, Fundamentals and Applications.

A.K. Peters Ltd., Natick.Touloukian, Y.S., 1967. Thermophysical Properties of High Temperature Materials.

IFI/Plenum, New York.Volkova, E.L., 2006. Modern magnesium-base deformable alloys and composite

materials (a review). Met. Sci. Heat Treat. 48, 473–478.Volovitch, P., Masse, J.E., Fabre, A., Barrallier, L., Saikaly, W., 2008. Microstructure

and corrosion resistance of magnesium alloy ZE41 with laser surface claddingby Al-Si powder. Surf. Coat. Technol. 202, 4901–4914.

Wang, A.A., Sircar, S., Mazumder, J., 1993. Laser cladding of Mg-Al alloys. J. Mater. Sci.28, 5113–5122.

Ye, H.Z., Liu, X.Y., 2004. Review of recent studies in magnesium matrix composites.J. Mater. Sci. 39, 6153–6171.

Yilbas, B.S., Mansoor, S.B., 2006. Laser evaporative heating of surface: simulationof flow field in the laser produced cavity. J. Phys. D: Appl. Phys. 39, 3863–3875.

alloy on magnesium. Mater. Lett. 61, 209–212.Yue, T.M., Lin, T., 2008. Laser cladding of Ni/Cu/Al functionally graded coating on

magnesium substrate. Surf. Coat. Technol. 202, 3043–3049.Yue, T.M., Su, Y.P., 2007. Laser multi-layer cladding of Zr65Al75Ni10Cu17.5 amorphous

alloy on magnesium substrates. J. Mater. Sci. 42, 6153–6160.