Impact Modeling of Thermally Sprayed Polymer Particles

Transcription

1 Imact Modeling of Thermally Srayed Polymer Particles Ivosevic, M., Cairncross, R. A., Knight, R., Philadelhia / USA International Thermal Sray Conference ITSC-2005 Basel, Switzerland, May 2005 Thermal sray has traditionally been used for deositing metallic, carbide and ceramic coatings, however, it has recently been found that the high kinetic energy of the High Velocity Oxy-Fuel (HVOF) thermal sray rocess also enables the solventless rocessing of high melt viscosity olymers, eliminating the need for harmful, volatile organic solvents. A rimarily goal of this work was to develo a knowledge base and imroved qualitative understanding of the imact behavior of olymeric articles srayed by the HVOF combustion sray rocess. Numerical models of article acceleration, heating and imact deformation during HVOF sraying of olymer articles have been develoed. A Volume-of-Fluid (VoF) comutational fluid mechanics ackage, Flow3D, was used to model the fluid mechanics and heat transfer during article imacts with a steel substrate. The radial temerature rofiles redicted using article acceleration and heat transfer models were used as initial conditions in Flow3D together with a temerature-deendent viscosity model to simulate olymer articles with a low temerature, high viscosity core and high temerature, lower viscosity surface. This aroach redicted deformed articles exhibiting a large, nearly hemisherical, core within a thin disk, and was consistent with exerimental observations of thermally srayed slats made using an otical microscoe. 1 Introduction The key advantages of using thermal sray rocesses for the deosition of olymers include: (i) solventless coating without the use of volatile organic comounds (VOCs); (ii) the ability to coat large objects under almost any environmental conditions; (iii) the ability to aly olymer coatings with high melt viscosity; and (iv) the ability to roduce ready-to-use coatings without the need for ost-deosition rocessing such as oven drying or curing, tyically needed for electrostatic owder coatings and solvent-based aints. The major disadvantages comared to those rocesses include: (i) lower deosition efficiency, (ii) lower quality surface finish and (iii) higher rocess comlexity, often with a narrow rocessing window defined by the olymer melting and degradation temeratures. Three thermal sray rocesses have reortedly been used for the deosition of olymers [1]: Conventional flame sraying. HVOF combustion sray. Plasma sray. Only a limited number of olymers srayed by the HVOF and lasma sray rocesses have been reorted and the commercial alications of HVOF and lasma srayed olymer coatings are still in the develoment stage [1]. The HVOF sraying of olymers has gained attention rimarily due to the significantly higher article seeds [u to 1,000 m/s] relative to flame sray [u to ~100 m/s]. This is an imortant advantage, esecially for the deosition of coatings with high melt viscosity, including high molecular weight olymers and olymer/ceramic comosites with high (>5 vol.%) ceramic reinforcement contents. 2 Background The thermal sray unit rocess is an individual article or drolet imacting onto a substrate to form a slat. Coating characteristics such as orosity, roughness, adhesive and cohesive strengths deend on the characteristics of these slats and how they bond to the substrate and to each other. As a result, many studies have examined drolet imact behavior in order to understand and imrove coating rocesses [2]. Virtually all of the ublished modeling of slatting behavior, however, has focused on metallic, ceramic or cermet articles. One of the rincial goals of this work was to develo a knowledge base and imroved qualitative understanding of the imact behavior of olymeric articles srayed by the HVOF combustion sray rocess. This may not only hel to imrove understanding of the relationshis between rocessing conditions, coating microstructures and coating roerties of HVOF srayed olymers but also the develoment of a new low temerature (< 500 C) thermal sray rocess caable of efficient and reliable deosition of olymers and olymer/ceramic comosites. As reviously reorted [3], the large differences in the roerties of Nylon 11 and zinc articles resulted in significantly different sreading behavior under similar thermal sray conditions, as may be reresented by the Peclet number (Pe). The Pe number for HVOF deosited olymer and metal drolets can be defined as a relative ratio of the drolets internal heat conduction time (t c ~ D 2 /α) and sreading time (t s ~ D/V) scales. Where (D), (V) and (α) are the diameter, imact velocity and thermal diffusivity of a article. The Pe number of a Zn drolet is in the range 1-10 while the Pe number of a Nylon 11 drolet is almost three orders of magnitude larger (Pe ~3,000), indicating that the sreading and cooling of Nylon 11 slats occur over two significantly different time scales. In addition, the substrate can be readily reheated to temeratures at or above the olymer melting temerature. This rovides conditions for ost-deosition melting of artially melted olymer slats that may enhance their wetting behavior and increase the adhesive strength of the coating. Postdeosition melting of the initial olymer slat layer on

2 a reheated substrate may rovide transition conditions for further coating build-u even if the HVOF srayed articles are only artially melted. 3 Mathematical Modeling 3.1 HVOF Gas Flow and Thermal Fields The combustion and exhaust gas characteristics inside the Jet-Kote II HVOF gun were determined based on calculations carried out by Dobbins et al., 2003 [4] using the same HVOF system as that utilized in this study. The adiabatic flame temerature reorted was calculated for a range of combustion chamber ressures and flame stoichiometries defined by a hydrogen-to-oxygen equivalence ratio Φ = (H 2 /O 2 )/(H 2 /O 2 ) stoichiometric using Chem-Sage thermochemical equilibrium software. At the equivalence ratio used in these exeriments, Φ = 0.83, the adiabatic flame temerature was ~2,830 ºC at a chamber ressure of 2.2 bar (220 kpa). These conditions, together with the gun internal geometry, were used to estimate a maximum jet velocity of V g * ~900 m/s (Mach ~ 0.6). An emirical correlation [5] for the axial mean velocity (V g ) of the HVOF jet with a range of validity between Mach number 0.3 and 1.4 was used as follows: V = V 1 ex α, for x x c, x 1 x c * g g > * V = V for x x, (1) g g, < c where (α = 0.85) is the gas velocity decay constant [5], (x) the axial distance along the gun barrel, and (x c ) the jet core length after the gun exit. The jet core length is a function of the nozzle exit diameter (D) and local Mach number (Ma) as defined by the emirical formula (2) [5]: x c = Ma 2 (2) D The water-cooled coer Jet-Kote II gun nozzle used in this work had a total length of mm of which the entry/transition region located between the combustion head and the nozzle itself was only 9.6 mm in length, the remainder (150 mm) was of constant internal diameter (6.35 mm). It was assumed that adiabatic, isentroic and frictionless fluid flow conditions existed within this ortion of the gun nozzle with a constant mean velocity and temerature. The mean axial gas temerature had the same functional form as the gas velocity (1) only with a larger exonential decay exonent (α = 1.35) as roosed by Tawfik et al. in 1997 [5]. 3.2 Particle Transort The velocity and temerature of articles during HVOF deosition are the key arameters defining their state and imact conditions and consequently the sreading behavior of imacting drolets. These were comuted using momentum and heat transfer equations for articles in the HVOF flow field. It is commonly acceted [6] that the drag force is the dominant force governing the movement of articles in an HVOF jet, so that article motion can be described by the following two ordinary differential equations: dv 1 m = C Dρ ga ( Vg V ) Vg V, V (0) = 0, dt 2 dx dt = V, x (0) = 0. (3) where (V ) is the article axial velocity, (A ) the article rojection area, (C D ) the drag coefficient, (ρ g ) the gas density and (x ) the article osition, calculated from the location where it enters the jet. Note that the relative velocity between article and gas (V g - V ) is multilied to its absolute value, which guaranties that a article is accelerated in a moving gas if its velocity is lower than the gas velocity and decelerated otherwise. The drag coefficient (C D ) is a function of the Reynolds number [7]: C ( Re ), for Re =, (4) Re D < This correlation was based on the assumtion that the articles were sherical, which is consistent with assumtions made in the heat transfer redictions and imact modeling. The Reynolds number for the relative gas flow around a article of diameter (D ) is defined as: Vg V Re = ρg D, (5) µ g where (µ g ) is the dynamic viscosity of the gas. The Biot number (Bi) of a article in an HVOF jet gives the ratio of internal and external heat transfer resistances. In other words, for a low Biot number the articles will be heated with negligible internal resistance, resulting in an almost uniform temerature distribution within the article. This is tyically true for most metallic and cermet materials used in thermal sraying (Bi < 0.1) and has led to a number of authors [4, 8-9] neglecting temerature gradients within the articles. In the case of olymers, the Biot number for a article in an HVOF jet is tyically much higher (Bi >

3 5) imlying that the most of articles are likely to develo large temerature gradients between the core and the surface. Accordingly, the equation describing the heat transfer from the gas to a single sherical article in sherical coordinates with aroriate initial and boundary conditions is of the form: T 1 T 2 C r k, 2 t r r r ρ = (6) T (r, t = 0) = o T dt dr dt = (R) g ) dr ( r 0) = 0, k ( r = R) = h ( T T, where (T ) is the article temerature, and (ρ ), (C), and (k ) are the density, heat caacity and thermal conductivity of the article, and (r) is a arameter which reresents the current article diameter taking values between (0) and the overall article diameter (R). The equations for momentum and heat transfer were solved by numerical integration using the Forward Euler method with a time ste small enough (10-7 s) that the local Reynolds number, gas velocity and temerature could be considered constant over each time ste. 3.3 Particle Imact A Volume of Fluid (VoF) method, develoed by Hirt and Nichols [10], namely FLOW-3D (version 9.0) was chosen to study the fluid mechanics and heat transfer during article imact with a substrate because of the comlex three-dimensional morhology of slats formed during thermal sraying. The initial imact conditions of an HVOF srayed article, such as article imact velocity and internal temerature rofile, were incororated into the FLOW-3D model from the momentum and heat transfer redictions, as described earlier. Molten Nylon 11 was modeled as a viscous shear thinning fluid with a temerature deendent viscosity using a Carreau model [11] in the form: µ = µ + µ 0 µ 1 n 2 2 ( 1 + λ γ& ) 2, (7) where (µ) is the olymer dynamic viscosity, (µ = 0) and (µ o = 13,000 Poise at T = 220 ºC) are infinite shear-rate and zero shear-rate viscosities, resectively, (λ = 1) is a time constant, (n = 0.7) is the ower-low exonent, and (γ = ,000 s -1 ) is the olymer shear rate. All coefficients were determined based on exerimental measurements on Nylon 11 material using the cone & late method, and the zero-shear-rate viscosity is a temeraturedeendent arameter. These results will be reorted searately. 4 Exeriments A semicrystalline Polyamide (Nylon 11) owder commercially available as Rilsan PA-11 French Natural ES D-60 (donated by Arkema) was used as the feedstock material in this work. As-received owder had a mean article size of 60 µm and corresonding article size distribution of -102/+26 µm. The melting and degradation temeratures of Nylon 11, as reorted by the manufacturer, were in the range o C and o C, resectively. Swie or slat tests involving single high seed [> 0.7 m/s] sray asses across room temerature glass slides at low owder feed rates [~2 g/min] were used to observe the morhology of individual slats. In addition, Nylon 11 articles were also deosited on a substrate reheated to 200 C in order to further understand the influence of substrate reheating on the deosition behavior of the Nylon 11 material. Slat tests were carried out using the Stellite Coatings, Inc. Jet-Kote II HVOF sray system using an O 2 /H 2 ratio of / m 3 /s (300/500 scfh) and sray distance of 200 mm. Slat morhologies srayed both with and without substrate reheating were analyzed using standard otical microscoy with olarized light (Olymus PMG-3 otical metallograh). In-flight article velocities at a distance of 100 mm from the nozzle exit were measured using the SrayWatch 2i system in conjunction with a diode laser illumination source, both rovided by Oseir Ltd. from Tamere, Finland. The article velocities were measured for four (4) different sray conditions varying the total gas flow rate at a constant oxygen: hydrogen ratio (Φ = 0.83). The four total gas flow rates used were 1.86, 2.23, 2.61 and 2.98 g/s. The tyical number of articles analyzed was in the range in all four cases. 5 Results and Discussion Most of the larger Nylon 11 slats srayed onto a room temerature substrate exhibited a characteristic fried-egg shae with a large nearly-hemisherical core in the center of a thin disk (Fig. 1). This shae indicated the existence of a large radial difference in flow roerties of the molten or nearly molten nylon drolets and largely unmelted core. In other words, the fried-egg shaed slats were formed by olymer articles having a large radial temerature rofile a low temerature, high viscosity core and a high temerature, low viscosity surface. On the other hand, nylon slats srayed onto a reheated substrate exhibited a flattened hemisherical shae (Fig. 2) likely due to ostdeosition flow activated by surface tension or/and residual stress after the initially fried-egg morhology slats were fully melted by the reheated substrate (Fig. 3). It was observed that ostdeosition flow of Nylon 11 slats occurred only when

4 the substrate was reheated to temeratures above ~185 C, which was consistent with the onset of melting of nylon s crystalline hase g/s at Φ = 0.83) did not follow the same trend of article velocity increase as was observed for the other three sray conditions. The highest O 2 + H 2 gas flow rates used in this exeriment were the only sray conditions corresonding to a sonic (~Mach 1) jet velocity, characterized by the aearance of exansion and comression ressure waves ( shock diamonds ). It was believed that the lower mean article velocity roduced at the highest O 2 + H 2 gas flow rate was the result of olymer article interactions with the shock structure. Fig. 1. Nylon 11 slats deosited onto a room temerature glass slide. Fig. 4. In-flight Nylon 11 article velocity as a function of O 2 + H 2 gas flow rate. Fig. 2. Nylon 11 slats deosited onto a reheated glass slide (200 C). Fig. 3. Nylon 11 imact sequence onto a reheated substrate, (I) artially melted article before imact, (II) fried-egg shaed slat, (III) ost-deosition flow of a fully molten drolet, (IV) drolet shrinkage during cooling. The results of in-flight article velocity measurements using the SrayWatch 2i system are shown in Fig. 4. Mean article velocities of 433, 479, 546 and 563 m/s were measured for the four gas flow rates investigated. Note that the highest O 2 + H 2 gas flow rate of scfh (a total gas mass flow rate of The redicted Nylon 11 article velocities as a function of article size are shown in Fig. 5. In general, olymer articles accelerated much faster relative to metallic or cermet articles in HVOF jets [4, 9]. This was due to the similar size but much lower densities of olymers ( g/cm 3 ) versus metals or cermets ( g/cm 3 ). For the same reason, however, the olymer articles will have lower inertia and their velocity will decay faster as they exit the gun nozzle than metals and cermets. One ossible way to maximize the imact velocity of olymer articles would be to reduce the sray distance. This aroach, however, has limited otential due to the otentially destructive effect of the HVOF jet on olymeric coatings at low gun surface seeds [12]. The redicted article velocity of 700 m/s for 60 µm articles at a 100 mm sray distance was significantly higher than the exerimentally measured article velocity ( m/s). This indicated that the gas flow and article acceleration models used require further otimization to rovide imroved agreement with exerimental measurements. The heat transfer from the combusting gas to Nylon 11 articles was comuted using the residence time of articles in the HVOF jet based on the redicted article velocity and the sray distance (Fig. 5). Predicted temerature rofiles within the Nylon 11 articles at a sray distance of 200 mm (8 in) at the

5 moment of imact with the substrate are shown in Fig. 6. These redictions were qualitatively consistent with the exerimental observations of Nylon 11 slats srayed onto a room temerature substrate. Larger articles (>45 µm) had unmelted or artially melted cores, which likely resulted in the formation of the fried-egg shae slats observed. flow of nylon slats on a reheated substrate is in rogress and will be reorted searately. Fig. 5. Predicted velocities of Nylon 11 articles in an HVOF jet (total O 2 + H 2 gas flow rate of 1.86 g/s at Φ = 0.83). Fig. 6. Predicted temerature rofiles within Nylon 11 articles immediately before imact with a substrate at 200 mm sray distance. The radial temerature rofiles redicted using the article acceleration and heat transfer models were used as the initial conditions in Flow3D for a temerature-deendent Carreau viscosity model. This generated articles with a low temerature (high viscosity) core and high temerature (low viscosity) surface. The redicted shaes of deformed articles exhibited a large nearly-hemisherical core in the center of a thin disk, which was consistent with exerimental observations of HVOF srayed slats (Fig. 7). Imrovement of the model to include multile article imacts and modeling of the ost-deosition Fig. 7. Simulated deformation of a Nylon 11 drolet with a radial temerature gradient and temeraturedeendent viscosity during imact. 6 Summary and Conclusions One of the rincial goals of this work was to develo a knowledge base and imroved qualitative understanding of the imact behavior of olymeric articles srayed by the HVOF combustion sray rocess. Mathematical models of article acceleration, heating and imact of Nylon 11 articles have been develoed. In general, olymer articles accelerated and decelerated much faster than metallic and cermet articles of similar size due to the much lower density of olymers. In addition, the high Biot number (Bi > 5) for olymer articles in an HVOF jet relative to metals indicated that most of the articles would likely develo a stee temerature gradient

6 between the article core and its surface. This was consistent with exerimental observations of thermally srayed slats, since most of the Nylon 11 slats srayed onto a room temerature substrate exhibited a fried-egg shae with a large nearly-hemisherical core in the center of a thin disk. The radial temerature rofiles redicted using article acceleration and heat transfer models were used as initial conditions in Flow3D with a temerature-deendent Carreau viscosity model; this generated articles with a low temerature (high viscosity) core and high temerature (low viscosity) surface. The redicted shaes of deformed articles also exhibited good qualitative agreement with exerimentally observed fried-egg shae slats. 7 Acknowledgements The authors would like to thank the National Science Foundation for roviding suort for this research under collaborative grant number DMI The views exressed in this aer do not necessarily reflect those of NSF. The authors also greatly areciate the assistance and hel of Dr. Thomas E. Twardowski, Mr. Dustin Doss, Mr. Varun Guta, Mr. Matthew Chalker, and Ms. Shannon Lafferty. The assistance of Oseir Ltd. with the article velocity measurements reorted here is also gratefully acknowledged. 8 References Overview, Key Engineering Materials, 197,.1-26, (2001). [7] Yang, X. and Eidelman, S., Numerical Analysis of a High-Velocity Oxygen-Fuel Thermal Sray System, Journal of Thermal Sray Technology, 5 (2), , (1996). [8] Pasandideh-Fard, M., Parshin, V., Chandra, S., and Mostaghimi, J., Slat Shaes in a Thermal Sray Coating Process: Simulations and Exeriments, Journal of Thermal Sray Technology, 11 (2), , (2002). [9] Li, M. and P. D. Christofides, Feedback Control of HVOF Thermal Sray Process Accounting for Powder Size Distribution, Journal of Thermal Sray Technology, 13 (1), , (2004). [10] Hirt, C. W. and Nichols, B. D., Volume of Fluid (VoF) Method for the Dynamics of Free Boundaries, J. of Comutational Physics, 39, , (1981). [11] Macosko, C. W., Rheology - Princiles, Measurements and Alications, Wiley-VCH, Inc., (1994). [12] Gawne D. T., Zhang T. and Bao Y., Heating Effect of Flame Imingement on Polymer Coatings, Proc. ITSC 2001, Eds. Berndt, C. C, Khor, K. A. and Lugscheider, E. F., Singaore, ASM International, Materials Park, OH, , (2001). [1] Petrovicova, E. and Schadler L. S., Thermal Sray of Polymers, International Materials Reviews, Vol. 47 (4), , (2002). [2] Fauchais, P., Fukomoto, M., Vardelle, A. and Vardelle, M., Knowledge Concerning Slat Formation: An Invited Review, Journal of Thermal Sray Technology, 13 (3), , (2004). [3] Ivosevic, M., Cairncross, R. A. and Knight, R., Heating and Imact Modeling of HVOF Srayed Polymer Particles, Proc International Thermal Sray Conference [ITSC-2004], DVS/IIW/ASM-TSS, Osaka, Jaan, May 10-12, (2004). [4] Dobbins, T. A., Knight, R. and Mayo, M. J., HVOF Thermal Sray Deosited Y 2 O 3 -Stabilized ZrO 2 Coatings for Thermal Barrier Alications, Journal of Thermal Sray Technology, 12 (2), , (2003). [5] Tawfik, H. H. and Zimmerman, F., Mathematical Modeling of the Gas and Powder Flow in HVOF Systems, Journal of Thermal Sray Technology, 6 (3), , (1997). [6] Cheng, D., Traaga, G., McKelling, J. W. and Lavernia, J. E., Mathematical Modeling of High Velocity Oxygen Fuel Thermal Sraying: An