Processing parameter effects on solution precursor plasma spray process spray patterns

Surface and Coatings Technology 183 (2004) 51–61

0257-8972/04/$ - see front matter � 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.surfcoat.2003.09.071

Processing parameter effects on solution precursor plasma spray processspray patterns

Liangde Xie , Xinqing Ma , Alper Ozturk , Eric H. Jordan , Nitin P. Padture , Baki M. Cetegen ,a b c c a c

Danny T. Xiao , Maurice Gell *b a,

Department of Metallurgy and Materials Engineering, Institute of Materials Science, University of Connecticut, Storrs, CT 06269-3136, USAa

Inframat Corporation, Farmington, CT 06032, USAb

Department of Mechanical Engineering, Institute of Materials Science, University of Connecticut, Storrs, CT 06269-3136, USAc

Received 4 October 2002; accepted in revised form 29 September 2003

Abstract

Solution precursor plasma spray (SPPS) is a promising process for making thermal barrier coatings (TBCs) with improveddurability. The improved durability of SPPS TBCs results from the novel microstructural features, (i) absence of splat boundaries,(ii) generation of through-thickness vertical cracks (iii) existence of uniformly distributed porosity. In this process, the coating isbuilt up by the overlapping and stacking of layers deposited in each pass of the plasma torch across the substrate. Informationconcerning the lamella is generally referred to as the spray pattern. The spray patterns from SPPS of 7wt.%Y O –ZrO precursor2 3 2

were studied via a fixed scan spray method, where the plasma torch repeatedly scans on a single line across the substrate. Thesurface of the resulting spray patterns can be divided into adherent deposits and powdery deposits that correspond to the hot andcold regions of the plasma jet, respectively. The penetration depth of the liquid into the plasma jet has a significant effect on theposition of the adherent deposits and the deposition efficiency. The structure of the adherent deposits across the section normal tothe motion of plasma torch is symmetric about its center. It becomes more porous and the thickness of the deposits is reducedtowards the edges. Implications of these results for understanding and improving the SPPS process are discussed.� 2003 Elsevier B.V. All rights reserved.

Keywords: Thermal barrier coatings; Plasma spray; Solution precursor plasma spray; Spray pattern

1. Introduction

The thermal spray process, especially the plasmaspray process, is a simple and economic process andhas been widely used to produce various coatings, suchas wear, corrosion resistant coatings and thermal barriercoatings (TBCs) w1–6x. Research on a new plasmaspray coating method, solution precursor plasma spray(SPPS), has been conducted by a number of investiga-tors w7–12x. Recently, this method has been developedto make highly durable TBC’s w10x. In contrast to theconventional plasma spray process, a solution precursorinstead of powder is used as feedstock in SPPS. TBCsproduced using SPPS process have the following novelmicrostructure w10x: (i) the absence of large splat bound-

*Corresponding author. Tel.: q1-860-486-3514; fax: q1-860-486-4745.

E-mail address: [email protected] (M. Gell).

aries (100 mm in length) that are usually the preferentiallocation for failure initiation and propagation, (ii)through-thickness vertical cracks providing coatingstrain tolerance, (iii) uniformly distributed porosityreducing the thermal conductivity and elastic modulusof the ceramic coating. These novel features of SPPSTBCs make the SPPS process promising for the produc-tion of TBCs with improved durability.

In the conventional plasma spray process, the coatingis formed by the overlapping and stacking of layersdeposited in each pass of the plasma torch across thesubstrate w13x. The layers are the result of the accumu-lation of many splats formed during the traverse ofplasma torch. Due to the spatial dispersion of coatingmaterial in the plasma jet, the temperature and velocityvariances of powders, the thickness w14x and microstruc-ture of the layer are not uniform across the sectionperpendicular to the motion of plasma torch. Also, as a

52 L. Xie et al. / Surface and Coatings Technology 183 (2004) 51–61

Fig. 1. Schematic of the fixed scan spray set-up.

Fig. 2. Schematic of the window-shield fixed scan spray set-up.

result of the interaction of plasma gas dynamics, particledynamics and heat transfer, the center of the depositlayer usually does not coincide with the axis of theplasma torch w15x. All of the above information aboutthe layer, i.e. thickness profile, microstructure varianceand position of the layer relative to the axis of theplasma torch, has been generally referred to as the spraypattern. The study of the spray pattern is very importantto the understanding of coating formation and optimi-zation of the process parameters w16x.

In SPPS, the coating is produced directly from asolution precursor and all physical and chemical reac-tions, such as evaporation, decomposition, crystallizationand coating formation, occur in a single step w7,8,17x.However, the coating is also built up by overlappingand stacking of layers deposited in each pass of theplasma torch. In this study, the effects of processingparameters on the spray patterns produced in the SPPSof TBCs are systematically investigated, and their impli-cations for understanding and improving of this processare discussed.

2. Experimental procedure

In order to investigate the spray patterns producedunder various processing conditions, a fixed scan sprayexperiment was designed, as shown schematically inFig. 1. In this experiment, the plasma torch moves onlyin one direction and the substrate is held stationary,which amplifies the spray pattern by multiple passesand avoids the overlapping of passes via fixing thevertical position of the plasma torch at the same time.The deposits were fabricated on 75=50=1.5 mm grit3

blasted (Al O grit, mesh size �320) stainless steel2 3

plates. The plasma torch used in these experiments wasMetco 9MB (Sulzer Metco, Westbury, NY) attached toa robotic arm. The investigated process parameters are(i) liquid feed rate (S), (ii) atomizing gas pressure (P)and (iii) argon (primary plasma gas) flow rate (Q).Each variable was evaluated at three levels, and therange of the studied parameters are 5–60 cm ymin for3

liquid feed rate, 0–2.76=10 Pa for atomizing gas5

pressure and 475–1575 cm ys for argon flow rate,3

respectively. For all experiments, the plasma power usedwas in the range of 35–45 kW, the secondary plasmagas was H . The solution precursor is atomized using a2

pressure atomizer (XAPR100, Bete Fog Nozzle Inc.,Greenfield, MA) before being injected into the plasmajet, and N is used as the atomizing gas. Measurements2

of the droplets using a phase Doppler particle analyzersystem show that the atomizer generates droplets withan average size of 38 mm and an average velocity of13 mys w17x. A 7wt.%Y O –ZrO (7YSZ) solution2 3 2

precursor provided by Inframat Corporation (Farming-ton, CT) was used as feedstock.

Loose particles were observed in the cross-section ofthe coatings, details will be presented in the results. Inorder to determine the origins of these loose particles inthe coating, a window-shield fixed scan spray experi-ment was designed, as illustrated schematically in Fig.2. In this experiment, a ceramic plate with a 20-mmdiameter hole in its center was placed in the front ofthe plasma torch and aligned with its centerline, so thatthe coating material traveling in the outer region of theplasma jet was shielded from the substrate.

The position of the adherent deposits, relative to theaxis of the plasma torch, was defined by measuring theplasma torch axis position on the substrate beforespraying and then marking it in the macrophotos takenusing a photographic unit (Nikon Macrophot�, Nikon,Japan). For the SEM investigation, the samples werecut normal to the plasma torch scanning direction,mounted with epoxy and prepared with standard metal-lographic procedures and then observed using an envi-ronmental scanning electron microscope (ESEM 2020,Philips Electron Optics, The Netherlands). From thesame samples, the thicknesses of the coatings weremeasured at an interval of 2 mm using an image analysissystem microGOP 2000 (ContextVisionin, Linkoping,Sweden) in conjunction with the optical microscope.The measured thickness data were then fitted to aGaussian function,

2w zB Ea xya0 1C Ff(x)s exp y0.5x |D Gay ~y 22pa2

53L. Xie et al. / Surface and Coatings Technology 183 (2004) 51–61

Fig. 3. A plot of Guassian function (a : the area underneath the curve,0

a : the center of the curve and a : the width of the curve at half1 2

maximum height).

Fig. 4. Typical macrophoto of the samples deposited using SPPS. (a) As-sprayed sample, (b) the same sample washed gently with water.

where x is the position, a is the area, a is the center0 1

and a is the width of the fitted curve at half maximum2

height, as shown in Fig. 3. The characteristics of thethickness profile, such as full width at half maximumheight (FWHM) (a ) and area (a ), were extracted2 0

accordingly from the fitted Gaussian function.

3. Experimental results

3.1. Effect of process variables on the position ofadherent deposits

Fig. 4a is a representative macrograph of the as-sprayed samples produced by fixed scan spray experi-ment. The coating can be divided into two regions: (i)adherent deposits and (ii) powdery deposits that arelocated both above and below the adherent deposits.Adherent deposits were identified as those that were notremoved by a stream of tap water, as shown in Fig. 4b.Between the powdery and adherent deposits, there is agap (the black region in Fig. 4a) where no deposit is

present. The adherent deposits are strongly bonded tothe substrate, while the powdery deposits are just looselyattached to the substrate.

During the spraying, the solution is injected down-wards into the plasma plume as shown in Fig. 1.Therefore, the position of adherent deposits relative tothe torch axis depends on the penetration of the liquiddroplets into the plasma jet. The deeper the solutionpenetrates into the plasma jet, the lower the adherentdeposits will be located relative to the plasma torchaxis. This trend is observed in the APS process w15x.Fig. 5 shows the change of the adherent deposits positionrelative to the torch axis at various operating conditions.It can be seen from Fig. 5a, when the liquid feed rate(S) increases from S to S , the position of the lower1 2

boundary of the adherent deposits does not change,however, the upper boundary moves upward by 7 mm.When the liquid feed rate increases further to S level,3

the position of the upper boundary of the adherentdeposits does not change further. However, the lowerboundary moves upward by 9 mm. When the atomizinggas pressure increases from P to P , as shown in Fig.1 2

5b, the lower boundary moves downward by 6 mm, andthe position of the upper boundary remains unchanged.With the further increase of the atomizing gas pressureto P , the upper boundary moves downward by 7 mm,3

and the lower boundary does not move downward anyfurther. For the lowest primary gas flow rate, as shownin Fig. 5c, the positions of both the upper and lowerboundaries are lower than the other cases. When theargon flow rate increases from Q to Q , there is little2 3

change of both the upper and lower boundaries.

3.2. Microstructure of deposits as a function of location

Fig. 6a is a representative macrophoto of samplesproduced in the fixed scan spray experiments, and Fig.6b–d are the top view of the illustrated regions of thesample. As shown in Fig. 6b–d, the powdery depositsconsist of loosely connected powders, while the structure


Fig. 5. Position change of the adherent deposits relative to the axis of plasma torch as a function of (a) liquid feed rate, (b) atomizing gaspressure, (c) argon flow rate.

of the adherent deposits is dense. This is consistent withthe observation in Fig. 4, that the powdery deposits canbe easily washed away with water.

Fig. 7a–f are the cross-section micrographs takennormal to the scanning direction of the plasma torchand each picture was taken at the locations shownschematically in Fig. 7g. It can be seen that the depositbecomes more porous and the thickness of the depositis reduced towards the edges of the deposit. This trendof microstructural changes is the same for both edgesof the deposits.

Fig. 8a and b are cross-section micrographs taken atthe center of the adherent deposits produced withoutand with the ceramic window in the front of the plasmatorch. Fig. 8a shows that many loose particles weretrapped in the adherent coating. When the precursordroplets traveling in the outer region of the plasma jetwere shielded from the substrate with the shield window,there were almost no loose particles in the coating (Fig.8b). The results indicate that the loose particles in thecoating originate from the precursor droplets travelingin the outer region of the plasma plume. The dense areain the deposits still forms when the precursor dropletstraveling in the outer region of the plasma jet wereshielded from reaching the substrate. In addition, theadherent deposits are lined up with the center of theplasma jet (Fig. 4). These results suggest that the densearea results from the solution fed into the inner part of

the plasma jet. This is verified by the results of othermodel spray experiments w18,19x.

3.3. Effect of process variables on the thickness profileof the adherent coating

In Figs. 9–11, (a–c) show the thickness profiles ofthe adherent deposits produced under various operationconditions. The wavy curves are the measured thickness.As indicated in these figures, the measured thicknessprofiles are fitted to a Gaussian function. For a fewcases, such as Fig. 10a and c, the consumption ofsolution precursor is not the same. In order to make upfor the difference in solution consumption in eachexperiment, the integrated area of the deposits is cor-rected on the assumption of linear accumulation ofdeposits with the number of passes. The correction madein this way is relatively small and the present studyprovides no information to develop a more precisenonlinear accumulation correction Figs. 9–11(d) showthe effects of each processing parameter on the charac-teristics of the fitted and corrected thickness profiles,i.e. FWHM and integrated area.

It can be seen from Fig. 9d that the FWHM of thefitted Gaussian function, decreases from 15.3 to 9.2 mmwhen the liquid flow rate increase from level S to S ,1 2

and it decreases to 8.3 mm when the solution feed rateincreases further to S . As shown in Fig. 10d, FWHM3


Fig. 6. Top view of the samples deposited using SPPS. (a) Macrophoto; (b and d) micrograph of the powdery deposits, (c) micrograph of theadherent deposits.

Fig. 7. Microstructure variance across the section normal to the motion of plasma torch showing that the deposit becomes more porous and thethickness of the deposit is reduced towards the edges of the deposit.

decreases from 11 to 9.2 mm when the atomizing gaspressure increases from P to P , and it becomes 8.51 2

mm when the atomizing gas pressure is P . In the range3

of argon flow rate investigated here, Fig. 11d indicatesthat FWHM decreases with the increase of the argonflow rate.

Another quantity we can extract from the thicknessprofiles is the area under the profiles. The product of

the cross-section area and the length of the depositsrepresent the volume of coating material deposited onthe substrate. Since the density of the adherent depositsis not significantly different and the lengths of thedeposits are the same, this area is a relative measure ofthe deposition efficiency. It can be seen from Fig. 9dand Fig. 10d that there is an optimum value for liquidfeed rate and atomizing gas pressure to achieve maxi-


Fig. 8. Microstructure of the center of the adherent deposits. (a)With-out window-shield, (b) with window-shield.

mum deposition efficiency. In the range tested for argonflow rate, the deposition efficiency increases with theincrease of argon flow rate, as shown in Fig. 11d.

4. Discussion

4.1. Deposition of precursor droplets fed into differentregion of the plasma plume

Semenov and Cetegen w20x using the present sprayingsystem and many other researchers w21–25x on similarsystems have shown that a large temperature variationexists in the plasma plume. The precursor droplets fedinto the plasma regions at different temperature willexperience different physical and chemical reactions.Fig. 12a is a schematic of the temperature distributionof the plasma jet. During the fixed scan spray, the first

coating material arriving at a given location in thesubstrate comes from the outer or colder region of theplasma jet, and is in the form of powdery deposits, asillustrated in Fig. 12b. The powdery deposits are thencovered over by the coating material from the hotterregion of the plasma jet for locations near the axis ofthe plasma torch and by the material from colder partof the plasma jet at all other locations. So, the adherentdeposits in the sample are a mixture of the coatingmaterial from both the hot and cold regions, and thepowdery deposits originate only from the precursor fedinto the cold region of the plasma jet.

Although the accumulation of loose powder is severalmillimeters thick when the plasma torch is held station-ary and all other operation parameters remains unchan-ged w19x, only a few microns of loose powder sticks tothe substrate. Meanwhile, the thickness of the adherentdeposits is in the range of several tens to severalhundreds microns, as shown in Figs. 9–11. The impli-cation of this result is that most of the coating materialcoming from the cold region of the plasma torch doesnot stay on the substrate by itself during the scanningof the plasma torch, while that from the hot region canadhere to the substrate and survive the subsequentscanning of the plasma torch. This observation is sup-ported by the results from the window-shield test whereadherent coating still forms when most of the precursordroplets traveling in the low temperature region of theplasma jet were shielded from the substrate. Therefore,the chemical and thermal status of the precursor dropletsis critical to ensure its bonding to the substrate orexisting coating.

4.2. Effect of solution droplets penetration depth on theposition and width of adherent deposits

During spraying, the liquid is injected downwardsinto the plasma jet, as shown in Fig. 1. Therefore, thepenetration depth of solution droplets into the plasmaplume (d) depends on both the momentum of droplets(M ) and momentum of the plasma gas (M ). Ms p s

decreases with the reduction of droplet velocity and sizedue to the evaporation of solvent, and M varies withp

the mass flow, velocity and temperature of plasma gas.According to the results obtained from the dropletvaporization model w26x (Appendix A) that consists ofthe droplet momentum equations as well as the massbalance, the relationships between penetration depth andliquid droplet size, injection velocity, plasma gas veloc-ity can be plotted as Fig. 13. As shown in the plots,higher injection velocity causes deeper penetration ofthe droplets. Similarly, larger droplets penetrate deeperthan the smaller ones, due to the higher momentum ofthose droplets. However, the effect of plasma gas veloc-ity is not as significant.


Fig. 9. Thickness profiles as a function of liquid feed rate. (a) S , (b) S , (c) S , (d) characteristics of the thickness profiles. (S -S -S ,1 2 3 1 2 3

measured: measured thickness profile, fitted: result of curve-fitting of the measured thickness profile with Gaussian function, corrected: fittedthickness profile with the correction of difference in solution precursor consumption.)

Fig. 10. Thickness profiles as a function of atomizing gas pressure. (a) P , (b) P , (c) P , (d) characteristics of the thickness profiles. (P -P -1 2 3 1 2

P , measured: measured thickness profile, fitted: result of curve-fitting of the measured thickness profile with Gaussian function, corrected: fitted3

thickness profile with the correction of difference in solution precursor consumption.)


Fig. 11. Thickness profiles as a function of argon flow rate. (a) Q , (b) Q , (c) Q , (d) characteristics of the thickness profiles. (Q -Q -Q ,1 2 3 1 2 3

measured: measured thickness profile, fitted: result of curve-fitting of the measured thickness profile with Gaussian function.)

Fig. 12. Schematic of coating build-up during fixed scan spray. (a)Temperature distribution in plasma jet, (b) end view of the substrateduring spraying.

The change of adherent deposits position, or thecenterline of the adherent deposits relative to the axisof the plasma torch at different processing parameterscan be explained by the penetration depth of the liquiddroplets (d). With the increasing of atomizing gaspressure, d increases due to the increase of liquidinjection velocity w26x, as shown in Fig. 13a, so thecenterline of adherent deposits moves downward, asshown in Fig. 5b. When the argon flow rate increasesfrom Q to Q , the adherent deposits moves upward as1 2

shown in Fig. 5c. The effect of plasma gas velocity dueto the variation of argon flow rate on the penetrationdepth of liquid droplets is small, as shown in Fig. 13b.The upward movement of the adherent deposits may beexplained by the enlargement of the plasma jet with theincrease of argon flow rate w25x, which increases theeffective cross-section area of the plasma jet that hasenough heat capacity to transform the precursor toadherent deposits. Further increase of argon flow ratefrom Q to Q may have little effect on the effective2 3

area of the plasma jet due to the cooling effect of higherargon flow rate. Therefore, the position of adherentdeposits does not change further, as shown in Fig. 5c.With increasing liquid feed rate, the penetration depthof liquid droplets changes with variation of liquid dropletsize and velocity, and with the reduction of plasma gastemperature, as shown in Fig. 13a. Therefore, the effecton the position of adherent deposits is difficult to predict.However, the upward movement of the adherent depositswith the increase of solution feed rate indicates that the

penetration depth of liquid droplets are decreasing,which is probably due to the dominant effect of dropletsize decrease with the increase of liquid feed rate.

Only a certain part of the plasma plume that hasenough heat capacity can heat the precursor to thechemical and thermal status required to assure its adhe-sion to the substrate. The cross-section area of this partof the plasma jet, A , depends on both the size of thep

plasma plume and temperature distribution in the plume.Thus, the width of the adherent coating or the positionof the adherent coating boundaries depends on both thecross-section area of the plasma plume with enough heatcapacity (A ) and the dispersion of solution precursorp

in this area. If precursor is only fed into a portion ofA , upper part for the case of under penetration andp

lower part for the case of over penetration, the width of


Fig. 13. Penetration depth of liquid droplets as a function of (a) liquidinjection velocity and droplet size, (b) plasma gas velocity.

adherent deposits will be narrower than that correspond-ing to the cases where solution precursor was fed to thewhole area of A . This can explain the change of thep

width of the adherent deposits in Fig. 5. For the casesof under or over penetration of precursor droplets intothe plasma jet, such as S , S , P , P , Q , the width of1 3 1 3 1

the adherent deposits are narrower than the correspond-ing cases, S , P and Q , where the solution precursor2 2 2

was fed to the whole area of A .p

The FWHM of the profiles always decreases mono-tonically with an increase of liquid feed rate, or liquidatomizing gas pressure, or plasma gas flow rate (Fig.9d, Fig. 10d and Fig. 11d). FWHM is a measure of theshape of the profile. A smaller FWHM indicates thatthe profile is narrower and the deposits are concentratedto a smaller area. The FWHM is dependent on thevariability of the vertical component of the momentumvector of the properly heated droplets that will bedeposited on the substrate as adherent deposits. A more

uniform distribution of the momentum of these properlyheated droplets will result in a smaller FWHM. Higherliquid feed rate results in smaller solution droplet size.Higher atomizing gas pressure leads to deeper penetra-tion of smaller droplets into the plasma jet. Both factorsresult in the proper heating of smaller solution dropletsand their deposition on the substrate as adherent depos-its. Since the difference in the absolute value of dropletmomentum resulting from the same velocity variation issmaller for smaller droplets, the deposition of smallersolution droplets thus result in smaller FWHM, whichmay explain the decrease of FWHM with the increaseof liquid feed rate and atomizing gas pressure (Fig. 9dand Fig. 10d). An increase of plasma gas flow rate willresult in the change of plasma gas temperature, velocityand density, which makes the understanding of its effecton FWHM very difficult. However, the decrease ofFWHM with an increase of plasma gas flow rate andthe accompanying increase of deposit volume (Fig. 11d)indicate that more precursor droplets are injected intothe high temperature region of the plasma jet and theirdeposition are concentrated in a smaller area.

4.3. Improving the deposition efficiency

Deposition efficiency is the ratio of coating materialdeposited on the substrate to that consumed totallyduring the spray process. From Figs. 9–11, it can beseen that the deposition efficiency is quite sensitive tothe operating conditions and can be related to thepenetration status of the solution precursor droplets intothe plasma jet, i.e. both under penetration and overpenetration of the liquid will decrease the depositionefficiency. So the key to improve deposition efficiencyis to feed as much precursor as possible into the properpart of the plasma jet by controlling the penetration ofsolution precursor into the plasma jet.

5. Summary and conclusions

Spray patterns from the SPPS process under variousoperating conditions are collected with fixed scan sprayexperiments. The position of the adherent deposits rel-ative to the axis of the plasma torch, the microstructuralvariance across the section normal to the motion ofplasma torch and the thickness profile of the produceddeposits were systematically studied.

The coating produced in the fixed scan spray experi-ment can be divided into adherent deposits and powderydeposits that correspond to the hot and cold regions ofthe plasma jet, respectively. The position of the adherentdeposits relative to the plasma torch axis varies with thespraying parameters, such as liquid feed rate, atomizinggas pressure and argon flow rate, and is related to thepenetration depth of the solution droplets into the plasmajet. Penetration of the solution precursor droplets into


the plasma jet depends on both the momentum of thesolution droplets and dynamics of the plasma gas. Thepenetration of the droplets has a significant effect onthe thickness profile of the lamella produced in one passand on the deposition efficiency. The structure of theadherent deposits across the section normal to the motionof plasma torch is symmetric about its center. It becomesmore porous and the thickness of the deposits is reducedtowards the edges. Powdery deposits in the samples andloose powder in the adherent deposits correspond to theprecursor traveling in the low temperature region of theplasma jet. Adherent deposits result from precursortraveling in both the hot and cold regions of the plasmajet. The coating material coming from the hot region ofthe plasma jet can adhere to the substrate and becomeadherent deposit by itself. The precursor traveling in thecold region may be incorporated into the adherentdeposits, however, it cannot form an adherent depositby itself.

Acknowledgments

We thank Drs Tania Bhatia and Peter Strutt for fruitfuldiscussions. This work was performed under the supportof US Office of Naval Research (Contract No. N000014-98-C0010 and Grant No. N000014-02-1-0171, moni-tored by Drs Lawrence Kabacoff and Steven Fishman).

Appendix A:

The penetration depth calculation presented in thisarticle is a part of the droplet vaporization model. Thismodel consists of the calculation of liquid dropletvaporization in the plasma gas stream. The liquid phasecalculations predict temperature and concentration vari-ations within the droplets injected into the hot plasmagas jet. The heat and mass transfer interactions with theplasma are obtained through the gas phase model. Thegas phase model contains the mass and energy balancesas well as the droplet momentum equations in the axialand transverse direction to determine the vaporizationof the droplet, its motion and trajectory in the plasmaas well as the velocity components and the heat trans-ferred into the droplet. The local properties of the plasmajet, namely the plasma temperature and velocity, areobtained from the experimental measurements w20x. Themeasured plasma gas data are utilized in the model todetermine the droplet vaporization.

The results presented in the article concern the droplettrajectory, more correctly, the penetration depth of thedroplet at the substrate location. To calculate the droplettrajectory, the droplet momentum equations in the axialand transverse directions are solved considering theaerodynamic drag as the principal force acting on thedroplets. The droplet momentum equation can be writtenas

™p dV ™d3r d syF .L d6 dt

Here, r is the density of the droplet, d is dropletL

diameter, V is the velocity vector, t is time and F isd d

the drag force. The drag force can be written in termsof the drag coefficient as

p 2 2F s r d C V .d g D8

In this equation, r is the plasma gas density andg

C is the drag coefficient. Since the relative velocityD

between the droplet and gas is important, the V term2

in the drag force is written as yNV yV N(V yV ) tog d g d

account for the direction of the drag force depending onthe relative velocity.

The governing equation then becomes

™B EdV 3 r C ™ ™ ™ ™C Fd g D ) )D Gs V yV V yV .g d g ddt 4 r dL

In this equation, V is the droplet velocity and V isd g

the local plasma gas velocity and it can be decomposedto the two droplet motion equations in the axial andtransverse directions. The drag coefficient of a spherewith Stefan flow correction for the surface blowingeffect is used as

24C sD Re 1qBŽ .M

The drag coefficient, C , depends on the ReynoldsD

number for the droplet, Res(r NV yV Nd)ym , and theg g d g

Spalding mass transfer number, B .M

References

w1x N.P. Padture, M. Gell, E.H. Jordan, Science 296 (2002) 280.w2x T. Valente, F.P. Galliano, Surf. Coat. Technol. 127 (2000) 86.w3x B.S. Mann, B. Prakash, Wear 240 (2000) 223.w4x E.H. Jordan, M. Gell, Y.H. Sohn, et al., Mater. Sci. Eng. A

301 (2001) 80.w5x M. Gell, E.H. Jordan, Y.H. Sohn, D. Goberman, L. Shaw, T.D.

Xiao, Surf. Coat. Technol. 146–147 (2001) 48.w6x W.J. Brindley, R.A. Miller, Adv. Mater. Proc. 136 (1989) 29.w7x J. Karthikeyan, C.C. Berndt, S. Reddy, J.Y. Wang, A.H. King,

H. Herman, J. Am. Ceram. Soc. 81 (1998) 121.w8x J. Karthikeyan, C.C. Berndt, J. Tikkanen, S. Reddy, H. Herman,

Mater. Sci. Eng. A 238 (1997) 275.w9x P.R. Strutt, B.H. Kear, R. Boland, US Patent No. 6025034,

2000.w10x N.P. Padture, K.W. Schlichting, T. Bhatia, et al., Acta Mater.

49 (2001) 2251.w11x S.D. Parukuttyamma, J. Margolis, H. Liu, et al., J. Am. Ceram.

Soc. 84 (2001) 1906.


w12x E. Bouyer, G. Schiller, M. Mueller, R.H. Henne, Appl. Organ-omet. Chem. 15 (2001) 833.

w13x R. McPherson, Plasma sprayed ceramic coatings, in: K.N.Strafford, et al. (Eds.), Surface Engineering Processes andApplications, Lancaster, PA, 1995, p. 3.

w14x H.C. Chen, E. Pfender, J. Heberlein, Plasma Chem. PlasmaProc. 17 (1997) 93.

w15x J.R. Fincke, W.D. Swank, D.C. Haggard, in: C.C. Berndt (Ed.),Thermal Spray: A United Forum for Scientific and Technolog-ical Advances, ASM International, Materials Park, OH, USA,1997.

w16x E. Pfender, Plasma Chem. Plasma Proc. 19 (1999) 1.w17x T. Bhatia, A. Ozturk, L. Xie, et al., J. Mater. Res. 17 (2002)

2363.w18x L. Xie, X. Ma, E.H. Jordan, N.P. Padture, D. Xiao, M. Gell, J.

Mater. Sci., in press.

w19x L. Xie, X. Ma, E.H. Jordan, N.P. Padture, T.D. Xiao, M. Gell,Mater. Sci. Eng. A 362 (1–2) (2003) 204.

w20x S. Semenov, B. Cetegen, J. Therm. Spray Technol. 10 (2001)326.

w21x M.K. Ferber, A.A. Wereszczak, M. Lance, J.A. Haynes, M.A.Antelo, J. Mater. Sci. 35 (2000) 2643.

w22x P. Fauchais, M. Vardelle, A. Vardelle, J.F. Coudert, Metall.Trans. B 20 (1989) 263.

w23x A. Kucuk, R.S. Lima, C.C. Berndt, J. Am. Ceram. Soc. 84(2001) 685.

w24x H.D. Steffens, T. Duda, J. Therm. Spray Technol. 9 (2000)235.

w25x S.J. Malmberg, Analysis of the plasma jet structure, particlemotion, and coating quality during DC plasma spraying, Ph.D.Thesis, University of Minnesota, 1994.

w26x A. Ozturk, B. Cetegen, unpublished research.

Documents

Processing parameter effects on solution precursor plasma spray process spray patterns