- Email: [email protected]
Development of process maps for plasma spray: case study for molybdenum
Materials Science and Engineering A348 (2003) 54 /66 www.elsevier.com/locate/msea
Development of process maps for plasma spray: case study for molybdenum S. Sampath a,*, X. Jiang a,1, A. Kulkarni a, J. Matejicek a,2, D.L. Gilmore b,3, R.A. Neiser b a
Department of Materials Science and Engineering, Center for Thermal Spray Research, State University of New York, Stony Brook, NY 11794-2275, USA b Thermal Spray Research Laboratory, Sandia National Laboratory, Albuquerque, NM, USA Received 11 April 2002; received in revised form 23 August 2002
Abstract A schematic representation referred to as ‘‘process maps’’ examines the role of process variables on the properties of plasmasprayed coatings. Process maps have been developed for air plasma spraying of molybdenum. Experimental work was done to investigate the importance of such spray parameters as gun current, primary gas flow, auxiliary gas flow, and powder carrier gas flow. In-flight particle temperatures and velocities were measured and diameters estimated in various areas of the spray plume. Empirical models were developed relating the input parameters to the in-flight particle characteristics. Molybdenum splats and coatings were produced at three distinct process conditions identified from the first-order process map experiments. In addition, substrate surface temperature during deposition was treated as a variable. Within the tested range, modulus, hardness and thermal conductivity increases with particle velocity, while oxygen content and porosity decreases. Increasing substrate deposition temperature resulted in dramatic improvement in coating thermal conductivity and modulus, while simultaneously increasing coating oxide content. Indentation reveals improved fracture resistance for the coatings prepared at higher substrate temperature. Residual stress was significantly affected by substrate temperature, although not to a great extent by particle conditions within the investigated parameter range. Coatings prepared at high substrate temperature with high-energy particles suffered considerably less damage in a wear test. The mechanisms behind these changes are discussed within the context relational maps, which have been proposed. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Process maps; Plasma spray; Thermal spray; Molybdenum; Residual stress
1. Introduction Thermal spraying is a well-established technology for the production of overlay protective coatings and is used extensively in industry servicing a range of components. Metallic and ceramic coatings allow engineering components to operate under extreme conditions of wear, corrosion and high-temperature exposure. More recently, sprayed coatings and multilayers are being considered for functional surfaces as well.
* Corresponding author 1 Presently at Caterpillar, Inc., Peoria, IL, USA. 2 Presently at Institute of Plasma Physics, Prague, Czech Republic. 3 Presently at CNA Corp., Alexandria, VA 22302, USA.
The use of thermal-sprayed coatings to date has been based on the concept of life extension of the engineering component, i.e. sprayed coatings are not considered as part of the component design. New demands arising particularly from the automotive and aerospace industries now require that the sprayed coatings are ‘‘prime reliant ’’ and that they be included during the design of the system. This requires considerable enhancement of the reliability and reproducibility of the coatings, as well as establishment of a knowledge base on their intrinsic materials properties and behavior. This has stimulated considerable research in the field. Thermal spraying is a highly complex deposition process with a large number of interrelated variables. Traditionally, satisfactory deposit quality is achieved
0921-5093/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 0 2 ) 0 0 6 4 2 - 1
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
through a design of experiment (DoE) approach. Numerous engineering parameters, such as power level, primary plasma gas type and flow rate, auxiliary plasma gas type and flow rate, stand-off distance, etc. have to be tried empirically and systematically through parameter matrix to determine the optimal parameters. The insight of the intermediate subprocesses and the corresponding mechanisms are largely unknown. The requirement of prime reliant coatings has created a need to develop strong scientific correlations among these complex parameters. This requires a concerted, integrated interdisciplinary approach, and this study demonstrates such an approach for a specific spray system and material. Process maps are a vehicle to establish these correlations based on sound scientific principles. From a scientific point of view, particle velocity (Vp) and temperature (Tp), substrate temperature (Ts) are the most important parameters, which determine the deposit build-up process and deposit properties. Particle temperature and velocity affect the deposit efficiency as well as the microstructure. Particle parameters affect almost every single subprocess in the deposition process, such as droplet spreading/solidification, droplet/substrate or previously deposited layer interaction, while substrate temperature has been found to have significant influence on splat morphology, deposit microstructure and properties [1 /3]. With an increase of substrate temperature, splat morphology changes from highly fragmented to contiguous, disk-like shape [1 /3], leading to enhanced deposit integrity, microstructure and properties [3,4].
55
ment, a good process map can enhance stability and reliability while also allowing process modifications to be successfully performed with a minimum of downtime, leading to improved economics. In a research environment, process maps allow one to validate theoretical models. The combination of experimental and theoretical data may allow interpolation and extrapolation to untried conditions. The knowledge gained in the development of a process map can be both fundamental and practical in nature. In this collaborative study, an effort was made to determine the relationships between spray parameters, in-flight particle characteristics, deposit microstructures and properties in a systematic manner. The aim of the study is to develop a more fundamental understanding of plasma /particle interactions, droplet /substrate interactions, deposit formation dynamics and microstructural development, as well as final deposit properties. The focus of this first part of this paper is the link between the processing parameters and the spray plume. The results of this part yield what is referred to as a firstorder process map for plasma spray. Based on these results, a set of process conditions was identified to generate splats and coatings to establish the linkages between process microstructures and properties. The correlations resulting from the second set of results yield a second-order process map . Fig. 1 illustrates the philosophy of such process maps for plasma spray. Ultimately, the combination of such maps provided an integrated set of understanding of the complex interactions in thermal spray and allow both fundamental understanding of the process as well as to provide tools for their utilization in the manufacturing sector.
1.1. Process maps A process map is an integrated set of relationships that link processing to properties and ultimately to performance. The map can consist of both experimental and computational data. Several types of linking relationships comprise a thermal spray process map. First, how do process conditions affect the particle plume? Experimental studies of this kind have become more common as particle diagnostics techniques have improved [5 /8]. Also, advances in the power of computational models have led to more theoretical studies of plasma/particle interactions [9,10]. Next, how does the particle plume interact with the substrate to produce the coating microstructure? And last, how does the microstructure behave when tested (i.e. properties and performance) and why? Studies linking particle characteristics to splat formation and coating microstructure and properties have also become more common in recent years [6,7,11 /13]. Process maps are useful because they allow one to more intelligently control the process in a non-research environment. For example, in a production environ-
2. Experimental 2.1. Torch set-up Plasma spraying was performed with a Praxair TAFA (formerly Miller Thermal) SG-100 plasma torch in a vertical orientation at Sandia National Laboratory. The Albuquerque, NM ambient atmospheric pressure for spraying was 83 kPa (12.2 psi). The torch hardware consisted of a 730 anode, a 720 cathode, and a 112 (straight) gas injector ring. The main plasma gas and the powder carrier gas were argon, the auxiliary plasma gas was helium, and the powder carrier gas was argon. Gas flow rates were measured using pressure transducers and jeweled critical orifices calibrated using a primary standard at Sandia Labs. The helium and argon were technical grade. A new anode/cathode set was used, and it was ‘‘broken in’’ prior to data collection by running for 30 min at 500 A, argon only. A straight powder feed tube with a 3.0 mm inner diameter was used, and powder injection was at 188
56
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
Fig. 1. Schematic overview of process maps for plasma spray.
forward of normal through a 1.8 mm diameter port. A plasma-densified fine size cut molybdenum powder (Osram Sylvania, #SD152) was used for the work presented in this paper. The volumetrically weighted powder size distribution as measured by a laser diffraction system (Model LS-100, Coulter Corp., Miami, FL) is given in Table 1. 2.2. Particle diagnostics Temperature, velocity, and diameter were measured for individual particles in the air plasma spray plume using a DPV-2000 diagnostics system (Tecnar Automation Ltd., St Hubert, Que.). The velocity is determined by time-of-flight between mask slits. The temperature is determined by two-color pyrometry. The diameter is calculated by integration of a pyrometry peak to give the radiated energy for a given particle. This is then used in conjunction with the temperature measurement to give the particle size. The depth of field and the diameter of the measurement region are both in the tenths of millimeter range. Further details of this system are available elsewhere [14]. The coefficient for the diameter calculations was calibrated by spraying with a different molybdenum powder which had a very narrow size cut. Fig. 2 compares the size distribution of the SD152 molybdenum powder before spraying with a typical size distribution for the in-flight particles.
Fig. 2. Molybdenum powder size distributions.
was fixed at 50 standard liters per minute (slpm) of argon and the powder feed rate was fixed at 0.18 g s 1 (1.5 lbs h1). A uniform-precision central-composite design was used to construct a 44-point experimental matrix in this three-dimensional parameter space, including an ‘‘outer’’ three-level full factorial cube (27 points), an ‘‘inner’’ two-level full factorial cube (eight points), and nine more repetitions of the central point (700 A, 18 slpm He, 2 slpm powder carrier gas) in order to distinguish the underlying measurement error from lack of fit in later empirical modeling. The parameter values at which particle measurements were collected are given in Table 2.
2.3. Design of experiments 2.4. Particle measurements The plasma torch parameters of importance to be investigated were torch current, auxiliary helium flow, and powder carrier gas flow (argon). During the experiments to determine the effects of these parameters on the particle characteristics, the main plasma gas flow
Having determined the parameter values at which particle measurements were to be made, the order in which measurements were randomized. The stand-off distance for the diagnostics sensors was fixed at 100 mm
Table 1 Molybdenum powder size distribution Volumetric mean (mm)
10% (mm)
25% (mm)
50% (mm)
75% (mm)
90% (mm)
2899
40
35
28
22
16
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66 Table 2 Torch parameters for DoEs particle measurements Parameter
Outer cube Inner cube
Current (A)
Helium (slpm)
Carrier gas (slpm)
500, 700, 900 580, 820
12, 18, 24 14.4, 21.6
1.0, 2.0, 3.0 1.4, 2.6
(4.0 in.) downstream of the exit plane of the torch. For each set of torch parameters, the sensors were then ‘‘auto-centered’’. That is, the sensors were positioned by an automated routine, which centered the measurement volume in the region of maximum particle flow. A map of particle flow in a planar cross-section of the plume is shown in Fig. 3. Note that in this figure, the central axis of the torch would pass through (0,0), and the powder injection tube would be parallel to the Y -axis at X /0. Thus, the measurements for this particular run would have been made at a position 6 mm (14 in:) from the central axis of the torch, in-line with the powder injection tube. 2.5. Fabrication of splats and deposits Based on the results of the first-order process maps, three conditions were selected with varying particle energies. These are nominally designated as low, medium and high. Splats were also collected on substrates nominally at three substrate temperatures */low: 115 8C, medium: 325 8C, high 465 8C. Details of the plasma and spraying condition are listed in Table 3. Characterization includes the observation and measurement of splats and the corresponding craters formed on the substrates. The procedure is as follows. First identification marks were made on the substrates by indentation so that the splats can be numbered and
Fig. 3. Distribution of particle flow rates within a plume cross-section at 100 mm stand-off. Data collected at ‘‘central’’ conditions (700 A, 18 slpm He, 2.0 slpm carrier gas).
57
located, subsequently splats were observed with optical microscopy and scanning electron microscopy (SEM) and dimensions were measured. Following this, the splats were etched away with saturated nitric acid and craters were characterized. The dimensions of splats and craters were measured with Zygo New View 200 noncontact surface profiler (a scanning white-light interferometer) [15]. Hardness measurement was carried out with a Buehler Micromet II microhardness tester using Vickers indentor and a 500 g load on the cross-section of the coating. High load indentation was carried out with a Mitutoyo AVK-C2 hardness tester at 20 kg load. Thermal conductivity measurements of the coatings were carried out on freestanding specimens by a laser flash technique, using Holometrix Thermal Properties Instrument. Open porosity was measured by mercury intrusion technique, using the Quantachrome Autoscan 33 porosimeter. Residual stresses were measured in 0.1 mm thick coatings on stainless steel. X-ray diffraction technique ‘‘sin2 c ’’ [16] was applied, using Siemens D500 diffractometer with Ni-filtered Cu radiation, reflection from (3 2 1) crystal planes with elastic constants E /313 GPa and n/0.31 and 10 sample tilts from c //528 to 528.
3. Results The first part of this paper examines the relationships between torch operating conditions and particle conditions as assessed by on-line in-flight diagnostics. As described in Section 2, a detailed evaluation of the particle conditions at the 44 DoEs was conducted. Based on the analysis of these results, a first-order process map relating particle velocity and temperature was generated and this served as the basis to produce splats and deposits for the second-order process map experiments. These results are summarized below. Fig. 4 displays particle temperature contours in a plume cross-section sprayed at the ‘‘central’’ conditions of the DoE. Comparing with Fig. 3, it can be seen that the region of hottest particles is shifted only slightly away from the region of maximum flow, and that all of the particles in this central area of the plume appear to be above the melting temperature of molybdenum (2610 8C). A quantitative estimate of the significance, or lack thereof, of particle segregation in the plume has been made by calculating the flow-weighted averages of particle temperature, velocity, and diameter at 121 points throughout the plume (a 10/10 mm2 with 1 mm point spacing). Table 4 shows that the standard deviations of the particle data are no more than 8% of the mean values. That is, although particle segregation exists, it is small, and the largest deviations in particle
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
58
Table 3 Plasma spraying parameters for splats and deposits Plasma power/parti- Current cle energy (A)
Plasma gas and flow rate
Carrier gas and flow Average particle temrate (lpm) perature (8C)
Average particle velo- Substrate temperacity (m s 1) ture (8C)
Low
540
Ar: 3.0
2780
130
115, 325a
Medium
700
Ar: 2.0
2980
153
115a, 325a
High
860
Ar: 50, He: 12 lpm Ar: 50, He: 18 lpm Ar: 50, He: 26 lpm
Ar: 1.0
3120
183
115a, 325a, 465
lpm: liter per minute. a Coatings were made at these Ts conditions.
Fig. 4. Distribution of particle temperatures within a plume crosssection at 100 mm stand-off and ‘‘central’’ conditions. Fig. 5. Mean particle temperatures and velocities at the various spray conditions are highly correlated. Table 4 Flow-weighted particle statistics for Figs. 3 and 4
Temperature Velocity Diameter
Mean9S.D.
S.D./mean (%)
2942969 8C 154911 m s 1 3393 mm
2.3 7.3 8.1
temperature and velocity occur at the edges of the plume where there are few particles present. 3.1. Correlations Fig. 5 indicates that for the 35 different spray conditions at which particle data were collected, the mean particle temperatures and velocities were highly correlated, as might be expected. This makes it difficult to separate the effects of particle kinetic energy and thermal energy on coating microstructure and properties. The mean particle temperature was found to be above the melting point of molybdenum for all spray conditions. Fig. 6 shows that, for a given run, particle velocity and diameter were also found to be correlated. The data fit fairly well to a curve derived from simple aerodynamic drag, with velocity being proportional to the inverse square root of the diameter. There is a signifi-
Fig. 6. Particle velocities for a given run agree with a simple aerodynamic drag model (V 8/1/d1/2). Data collected at ‘‘central’’ conditions (700 A, 18 slpm He, 2.0 slpm carrier gas) at spray pattern center (maximum particle flux).
cant amount of scatter in the data about this curve, presumably due to the statistical distribution of particle injection trajectories. 3.2. Empirical model The data from Fig. 5 were used to create an empirical model of the effects of spray parameters on particle
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
59
temperature and velocity (little effect on particle diameter was observed). The parameters (current, helium flow, powder carrier gas flow) were first normalized for the ranges investigated so that the relative importance of each term in the model could be accurately gauged. The particle characteristics were then fit to a quadratic surface in the three-dimensional parameter space: Tp =Vp C0 C1currentC2helium C3carrier gasC4currentcurrent Fig. 8. Pareto plot of parameter effects on particle velocity.
C5currenthelium C6currentcarrier gas C7heliumhelium C8heliumcarrier gas C9carrier gascarrier gas;
(1)
where all parameters are normalized and the coefficients are as given in Figs. 7 and 8. For example, referring to Eq. (1) and Fig. 7, for spray conditions of 800 A, 21 slpm helium, and 2.5 slpm powder carrier gas, the model would predict a particle temperature of 2999 8C. All the parameters in this example would have normalized values of 0.50 (e.g. 800 A is half-way between the ‘‘central’’ condition of 700 A and the highest current used in the DoE, 900 A). 3.3. Relative importance of parameters Figs. 7 and 8 show that the plasma current is the dominant influence on the particle temperature and velocity, as might be expected. In the range of values, which were studied, increasing powder carrier gas flow decreased both temperature and velocity, indicating that this system is fairly sensitive to the trajectory of the injected particles. The secondary gases impart good heat transfer properties thereby increasing plasma voltage and energy. However, helium flow has a surprisingly small effect on particle temperature*/an increase of 6 slpm only results in a 9 8C rise. Fig. 9 plots the predicted values for particle temperature and velocity from the empirical model given above. Values were calculated over the full range of parameter space investigated: 500/900 A, 12 /24 slpm helium, 1.0 /
Fig. 9. Predicted and measured values for mean particle temperature and velocity.
3.0 slpm carrier gas. For comparison, the 44 experimentally measured data points are also shown. As noted previously, temperature and velocity are highly correlated. However, the predicted values show that by properly adjusting the spray parameters, one may vary temperature and velocity independently within small ranges, more so at low values than at high values. 3.4. Test and use of model In order to test the model, new experimental data were taken at previously untested points in parameter space. Some of these new points were within the limits of the original DoE cube (testing interpolation), and some lay outside those limits (testing extrapolation). The new measured values were found to agree with predicted values within 2% on average for temperature, and within 3% on average for velocity. In general, interpolated predictions were found to be more accurate than extrapolated ones, as might be expected. 3.5. Fabrication of splats and deposits
Fig. 7. Pareto plot of parameter effects on particle temperature.
Once process maps linking the torch parameters to the particle characteristics were completed, conditions were chosen for the synthesis of samples. These conditions are indicated in Table 3 and illustrated in Fig. 10.
60
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
Fig. 10. Processing condition for process map samples. Particle energies are 350 J mol 1 (low), 540 J mol1 (medium) and 700 J mol 1 (high) and substrate temperatures vary between 115 8C (low) and 325 8C (high).
As in the DoE work, the torch axis was vertical, and the torch /substrate stand-off distance was 100 mm (4.0 in.). Substrates were mounted eight at a time in a carouseltype fixture, roughly 150 mm (6 in.) in diameter and rotating about a horizontal axis. Thus, a given substrate was normal to the torch axis every one-eighth of a revolution. As substrate temperature has been found to have a large influence on coating microstructures and properties [17,18], the surface temperatures (Ts) of the samples were monitored and controlled by an air-cooling feedback system. This consisted of an infrared pyrometer which measured the temperature of a given substrate (or top surface of the deposit) at a position one-quarter of a revolution after it had passed under the spray plume. This temperature data were then used to control the pressure of the cooling air jet, which was positioned 1808 opposite to the torch. With this system, it was typically possible to hold a desired temperature within 5 8C. However, the range of substrate temperatures achievable were limited in some cases by the available cooling air flow and the particular heat input for a given set of plasma conditions. For a given set of conditions, three types of deposit samples were sprayed: splats (to study the effect of inflight particle characteristics and substrate temperature on morphology), thin coatings (for wear and residual stress measurements), and thick coatings (for metallography, porosity and thermal conductivity measurements). The specimen configurations are schematically illustrated in Fig. 10. Splats were produced by passing the torch once over the substrate carousel (rotating at 240 rpm) at a traverse rate of 100 mm s 1 (4.0 in. s 1) and a powder feed rate of approximately 0.05 g s 1 (0.4 lbs h1). Thin coatings, nominally 100 mm (0.004 in.) thick, were produced by passing the torch over the substrates (150 rpm) about 50 times at a traverse rate of 25 mm s 1 (1.0 in. s 1) and a powder feed rate of approximately 0.18 g s 1 (1.5 lbs h1). Thick coatings,
nominally 1.3 mm (0.05 in.) thick, were produced with about 250 passes at 7.5 mm s 1 (0.3 in. s1), a feed rate of 0.18 g s 1 (1.5 lbs h 1), and a carousel rotational speed of 180 rpm. Particle temperature, velocity, and diameter were measured for every spray run to insure that they were within the target ranges. 3.6. Droplet/substrate interaction and the establishment of adhesion It was found that molybdenum splat morphology is sensitive to particle conditions and substrate temperature. At low substrate temperature (Ts), all splats show highly splashed morphology with a fragmented core structure and debris scattered radially. With an increase in substrate temperatures and decrease of particle energy, contiguous splat is more readily formed as schematically shown in Fig. 11.
Fig. 11. Process map illustrating splat morphologies at various processing conditions. Particle energies are 350 J mol 1 (low), 540 J mol 1 (medium) and 700 J mol1 (high) and substrate temperatures vary between 115 8C (low) and 325 8C (high).
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
It is found that substrate is significantly melted and deformed upon the impact of droplet, which leads to the formation of flower-like splats and craters. On average, only about 36 /53% of the areas covered by splats were in good metallurgical/mechanical contact with substrate, as indicated by substrate alloying and/or deformation, with higher particle energy/higher Ts combination shows higher effective contact area ratio. Crater volume can reach up to 35% of the impacting droplet volume, indicating very severe substrate deformation [19].
3.7. Coating microstructure and integrity Fig. 12 shows the cross-section of coatings prepared at Ts /115 8C at three particle energy conditions. All of them display lamellar microstructure. With the increase of particle energy, the lamellae in the coatings become slightly thinner due to the higher particle velocity. The number of interfaces is larger for higher particle energy for a given thickness. It appears that porosity is slightly higher in the low particle energy coating. Inter-lamellar pores can be discerned clearly in all three coatings. Coatings produced at higher Ts show similar microstructure. Indentation method was used to reveal possible differences in coating mechanical response originated from different microstructure at various PE and Ts conditions. The impressions and surrounding coating structures are shown in Fig. 13. At the load of 500 g, numerous inter-lamellar cracks can be found around the impression for all the coatings produced at Ts /115 8C regardless of the particle energy level. Meanwhile, for the coatings produced at higher Ts, no cracks were observed around the impression. It is noticed that interlamellae boundaries are more readily seen in the higher Ts coatings. This feature is probably due to the higher oxide level on the interface associated with higher Ts coatings (see discussion in the oxidation section). With the increase of load to up to 20 kg, low Ts coatings show similar cracking pattern and higher Ts coatings show long cracking initiated from the tips of indentation diagonal and propagating parallel to the deposit/sub-
61
strate interface along lamellae/lamellae boundaries as shown in Fig. 14 (medium PE). Initiation of cracking is determined by the local stress level and the coating fracture toughness. The existence of many short cracks around the low Ts coatings indentations indicates evidently that the adhesion between lamellae and lamellae is much weaker compared with coatings produced at higher Ts. The stress is relaxed even when it is small and therefore no large stress built up near the indentation tip. In the case of higher Ts coating, stress is built-up and cracks initiated at the indentation tip where maximum stress exists. This observation is consistent with our observation before in another study, where, high Ts molybdenum coating showed trans-lamellae fracture feature as compared with inter-lamellae mode in the low Ts deposit [3]. The fact that cracks are parallel to the deposit/substrate interface is a manifestation of the anisotropy in deposit structure and property. 3.8. Residual stress The effects of deposition temperature and particle energy on residual stress are presented in Table 5 and Fig. 15. As can be seen from Fig. 15a, deposition temperature has quite a dramatic effect on the coating residual stress. When the temperature increases from 120 to 280 8C, the stress shifts from tensile to compressive and becomes even more compressive above 300 8C. This trend can be explained by increasing contribution of thermal stress, which is proportional to the difference in coefficients of thermal expansion (CTE) of the substrate and the coating, temperature drop during post-deposition cooling and coating modulus. Since the coating’s CTE is lower than the substrate’s, the thermal stress is compressive and its magnitude increases with deposition temperature due to larger temperature drop after the deposition. Increasing modulus with deposition temperature [3] can explain the increase in slope in Fig. 15a. The effect of particle energy is rather negligible, as shown on Fig. 15b. Only a small trend of decreasing stress with particle energy was observed at lower deposition temperature; however,
Fig. 12. Cross-section of coating made at three particle energy conditions at Ts /115 8C etchant: Murakami’s. (a) low PE (350 J mol 1), (b) medium PE (540 J mol1), and (c) high PE (700 J mol 1).
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
62
Fig. 13. Indentation mark and cracking pattern on coatings produced at different particle energies (PE) and Ts conditions, load 500 g. (a) Low PE (350 J mol 1), (b) medium PE (540 J mol1), and (c) high PE (700 J mol 1).
Table 5 Residual stresses for coatings produced at various spray conditions Particle energy
Ts (8C)
Low
120 275
13
20
Medium
120 285
12 92
16 15
High
120 275 330
10 92 331
19 18 18
Stress (MPa)
S.D.
the differences were smaller than experimental error. The particle energy effects may become more significant in thick coatings, where the associated changes in splat/ coating formation will affect the microstructure and thus mechanical properties. Compressive stress in the coating may be beneficial for its mechanical properties, since it would close the existing cracks and suppress crack propagation, but too
high magnitude can lead to delamination at the coating/ substrate interface. The stress state and mechanical properties can be easily controlled by the deposition temperature and therefore can be tailored to specific application. 3.9. Oxygen content Oxygen content of coatings produced at various conditions is plotted in Fig. 16. It is clear that at 115 8C, with an increase of particle velocity, oxygen content decreases. For the same particle velocity, coating produced at higher Ts show higher oxygen content. This result corroborates that both in-flight particle oxidation and post-depositing surface oxidation contribute to the overall coating oxidation. It is shown through modeling that in-flight oxidation is primarily controlled by the residence time of molten particle in the oxygen containing flame [20]. Since the nozzle/substrate distance is fixed in all the conditions, high velocity particles will have low oxygen content. At
Fig. 14. Indentation mark and crack pattern on coatings produced at two Ts conditions, particle energy level: medium (540 J mol 1); load: 20 kg.
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
63
Fig. 17. Microhardness of coatings produced on various particle energy and substrate conditions.
Fig. 15. (a) Residual stress vs. deposition temperature for different particle energies. (b) Residual stress vs. particle energy for different deposition temperatures. Fig. 18. Mercury intrusion porosity of coatings produced on various particle energy and substrate conditions.
3.10. Hardness and porosity As shown in Figs. 17 and 18, with the increase of particle kinetic energy (calculated from particle velocity and size data), coating porosity decreases significantly and coating hardness increases slightly for the two Ts investigated. Substrate temperature shows a dramatic effect on coating hardness. With an increase of Ts from 115 to 325 8C, hardness increases by more than 20%. But the porosity change is not as dramatic. Fig. 16. Oxygen content of coatings produced on various particle energy and substrate conditions.
3.11. Thermal conductivity Table 6 lists the thermal conductivity data for coatings made at various conditions. These values are only
higher Ts, post-depositing surface oxidation is much more significant, and so the overall oxygen content is increased. The reason for the higher oxygen content at 180 m s1 coating as compared with 160 m s 1 coating at high Ts is most likely associated with marginally increased Ts for 183 m s 1 coating (325 8C) compared with that for the 153 m s 1 coating (310 8C).
Table 6 Thermal conductivity of the coatings Thermal conductivity (W m 1 K 1)
Ts 115 8C
Ts 325 8C
Medium PE High PE
13.0 16.6
46.5 45.3
64
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
about 10/30% of that for sintered molybdenum, associated with the existence of many inter-lamellae interfaces and pores. It appears that thermal conductivity is not significantly affected by particle energy but very sensitive to Ts. Thermal conductivity tripled with the increase of substrate temperature from 115 to 325 8C. Thermal conductivity is an indication of inter-lamellae contact quality. The dramatic increase of thermal conductivity reveals significant enhancement of interlamellae bonding and adhesion.
4. Discussion 4.1. Effect of particle conditions and substrate temperature
reveal detectable amount of MoO3 inside the coatings [25]. Perhaps the low melting point oxide melts and dissolves into the impinging droplet, assisting to form good metallurgical bonding between splats and improve thermal conductivity. If the oxide exists as a solid film at the interface, the coating thermal conductivity would have been reduced. Similar coating property enhancement at higher Ts was reported in partially stabilized zirconia deposits [4]. This suggests mechanism (3) is unlikely to be the only mechanism. In reality, perhaps all three mechanisms are active. However, the most plausible explanation is based on #1 the role of adsorbates and condensates. Recent works from some of the authors have shown this to be a rather dramatic effect [26]. 4.2. Synthesis second-order process maps
The results of the first-order process maps indicate that for a given plasma spray nozzle, the particle velocity and particle temperature are highly correlated. Consequently, parameters that cause an increase in one of the variable generally results in the increase in the second variable. In this work, the coupled variable of temperature and velocity is treated as particle energy and the combined effect of this parameter is addressed through integration into the second-order process maps. The particle energy has the effect of enhancing the density of the deposit and consequently an increase in elastic modulus. For the explored range of particle energy space, there is almost no effect on residual stresses. The particle oxidation is reduced in spite of increase in particle energy, given the reduced dwell time of the particles. It is evident that substrate temperature has a far more influencing effect on coating microstructure, hardness, residual stress and inter-lamellae contact than particle energy within the investigated parameter ranges. Coatings produced at higher Ts also show lower coefficient of friction and higher scratch resistance [9]. Although, increasing particle energies show effect in reducing coating porosity and increasing hardness, the improvement is not dramatic. The mechanism of improvement in inter-lamellae adhesion at higher Ts has been a subject of considerable analysis in recent years [21 /23]. Our current understanding suggests that there are three possibilities: (1) reduced adsorbates/condensates at higher substrate temperatures; (2) less trapped air in the pores due to lower air density at higher temperature, lower air density results in less severe disturbance to droplet spreading may lead to better contact; and (3) thicker splat surface oxide layer (presumably MoO3 as suggested in literature [24]) as evidenced by the higher oxygen content may help to improve the contact quality. XPS study shows trace of MoO3 on the as-sprayed coating surface, but Xray and electron diffraction of similar coatings did not
The second-order process map links the effect of key process variables on the microstructure and properties of the sprayed materials. The goal here is to identify the sensitivities of these variables on the properties of the deposit while simultaneously generating regime maps which will not only provide a fundamental understanding of intrinsic deposit properties but also enable their incorporation into design. It is envisioned that such maps will provide designers means to consider the coating into the system and also allow manipulation of the properties to achieve specific goals. Fig. 19 provides a schematic representation of one such process map. In this diagram, three attributes of the deposit are identified with respect to the processing conditions and based on their sensitivities. The lines with arrows depict a trend in the property within the process operational regime and slope suggests the sensitivity of the parameter. The dotted line represents a zero stress line where the tensile residual stress generated during the quenching of the splat is matched by the thermal mismatch stress for the steel /Mo system
Fig. 19. Illustrating 2nd-order process map for plasma spraying of molybdenum on steel substrates.
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
at the deposition temperature. In addition, regimes of lower oxide content in the coating are also depicted in the map. The maps provide important information about the process and the resulting deposit. The key ones are identified as follows: . Increasing particle energy is a major driver for reducing porosity in the material. Increasing deposition temperature can also reduce porosity but this effect is less sensitive to that of particle energy. The elastic modulus of the deposit correlates with that of the overall porosity of the coating. . Increasing particle energy reduces the dwell time of the particles within the plasma and the associated entrainment field and reduces the in-flight oxidative processes. However, increasing deposition temperature increases the oxide content in the deposit, arising from surface oxidative phenomena. . Increasing particle energy has no effect on residual stresses in the deposit consistent with the quenching stress based residual stress in the system. At higher deposition temperatures, the thermal stress field is superimposed on the quenching stress. This can lead to either tensile or compressive overall residual stress depending on the mismatch in the substrate-coating system. . Through thickness thermal conductivity of the deposit is more strongly affected by the substrate temperature rather than an increase in overall porosity. This is due to the improvement in wetting and associated contact between lamellae within the deposit.
5. Conclusions Thermal spraying is often very sensitive to process details and substrate condition and this has a profound effect on the properties and ultimately the performance of the sprayed coating. Process maps offer a means to relate process conditions to the critical properties of the sprayed coatings in a systematic and intelligent manner. Diagrams relating torch parameters to the thermal and kinetic energy of sprayed particles can be constructed rapidly, typically in a few days to a week. These diagrams can be used to intelligently screen for processing conditions that are likely to produce a wide range of coating microstructures. This screening reduces the need for time-consuming characterization of numerous samples, frequently requiring weeks to months. In this paper, air plasma spray of molybdenum on steel substrate was studied. An empirical model was developed from experimental data to link the torchinput parameters to the properties of the spray plume. The mean velocity and temperature of the in-flight
65
particles were found to be strongly correlated. Applied current was found to have the dominant influence on particle characteristics. The trends observed from these experiments can be constructed in the form of process maps, which illustrate the correlations and identify the sensitivities of the parameters to the output variables. Linking these process variables to the deposit formation and dynamics and ultimately the deposit properties allows an integrated understanding of the process to properties. It was found that there is a strong mechanical/thermal interaction between droplet and substrate, which builds up the coating/substrate adhesion. Hardness, thermal conductivity increase, oxygen content and porosity decreases with increase of particle velocity. Increasing deposition temperature resulted in dramatic improvement in coating thermal conductivity and hardness as well as increase in coating oxygen content. The residual stress in the coatings changes from tensile to compressive as the deposition temperature increases for the steel /molybdenum system, which makes this parameter an effective means of controlling the stress level. In the investigated parameter range, substrate temperature has a larger effect on coating property as compared with particle energy. A type of process map has been generated based on this systematic study. Although this has been demonstrated for a given material /substrate combination, these concepts can be generalized. This provides a means enhance fundamental understanding of the complex interactions during thermal spray processing and illustrates sensitivities of certain process variables to critical deposit properties. The framework also allows visualization of process regimes to achieve desired coating properties. Such strategies will allow more expansive and intelligent utilization of thermal spray coatings.
Acknowledgements This research was supported by the MRSEC program of the National Science Foundation under award DMR 9632570 and DMR 00890021. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under Contract DE-AC04-94AL85000. The authors would also like to acknowledge the role of David E. Beatty in the collection of particle data and the production of test samples.
References [1] L. Bianchi, F. Blein, P. Lucchese, M. Vardelle, A. Vardelle, P. Fauchais, in: C. Berndt, S. Sampath (Eds.), Thermal Spray
66
[2]
[3] [4] [5]
[6]
[7]
[8]
[9] [10] [11]
[12]
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66 Industrial Applications, ASM International, Metals Park, OH, 1994, p. 569. M. Fukumoto, S. Katoh, I. Okane, in: A. Ohmori (Ed.), Proceedings of the Integrated Thermal Spray Conference, Kobe, Japan, Japan High Temperature Society, 1995, pp. 353 /358. X. Jiang, J. Matejicek, S. Sampath, Mater. Sci. Eng. A 272 (1999) 189. S. Sampath, X.Y. Jiang, J. Matejicek, Mater. Sci. Eng. A 272 (1999) 194. J.R. Fincke, W.D. Swank, D.C. Haggard, T.M. Demeny, S.M. Pandit, A.R. Kashani, in: C.C Berndt, S. Sampath (Eds.), Advances in Thermal Spray Science and Technology, ASM International, Materials Park, OH, 1995, pp. 117 /122. R.N. Wright, J.R. Fincke, W.D. Swank, D.C. Haggard, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 511 /516. M. Prystay, P. Gougeon, C. Moreau, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 517 /523. W.D. Swank, J.R. Fincke, D.C. Haggard, S. Sampath, W. Smith, in: C.C. Berndt (Ed.), Thermal Spray: A United Forum for Scientific and Technological Advances, ASM International, Materials Park, OH, 1997, pp. 451 /458. J.D. Ramshaw, C.H. Chang, Plasma Chem. Plasma Process. 12 (1993) 299. Y.P. Wan, V. Prasad, G.X. Wang, S. Sampath, J.R. Fincke, Int. J. Heat Mass Transfer 121 (1999) 559. A.C. Le´ger, M. Vardelle, A. Vardelle, P. Fauchais, S. Sampath, C.C. Berndt, H. Herman, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 623 /628. S. Sampath, J. Matejicek, C.C. Berndt, H. Herman, A.C. Le´ger, M. Vardelle, A. Vardelle, P. Fauchais, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 629 /636.
[13] M. Vardelle, P. Fauchais, A. Vardelle, A.C. Le´ger, in: C.C. Berndt (Ed.), Thermal Spray: A United Forum for Scientific and Technological Advances, ASM International, Materials Park, OH, 1997, pp. 535 /542. [14] C. Moreau, P. Gougeon, M. Lamontagne, V. Lacasse, G. Vaudreuil, P. Cielo, in: C.C. Berndt, S. Sampath (Eds.), Thermal Spray Industrial Applications, ASM International, Materials Park, OH, 1994, pp. 431 /437. [15] P. De Groot, J. Mod. Opt. 42 (1995) 389. [16] J. Matejicek, S. Sampath, J. Dubsky, J. Therm. Spray Technol. 7 (1998) 489. [17] Haddadi, F. Nardou, A. Grimaud, P. Fauchais, in: C.C. Berndt, S. Sampath (Eds.), Advances in Thermal Spray Science and Technology, ASM International, Materials Park, OH, 1995, pp. 249 /254. [18] J. Matijicek, S. Sampath, H. Herman, in: C. Coddet (Ed.), Thermal Spray: Meeting the Challenges of the 21st Century, ASM International, Materials Park, OH, 1998, pp. 419 /424. [19] X.Y. Jiang, Y. Wan, X.Y. Wang, H. Zhang, R. Goswami, H. Herman, S. Sampath, in: C.C. Berndt (Ed.), Thermal Spray, Surface Engineering via Applied Research, ASM International, Materials Park, OH, 2000, pp. 729 /736. [20] Y. Wan, X. Jiang, H. Zhang, S. Sampath, V. Prasad, J. Fincke, Metall. Mater. Trans. B 32 (2001) 475. [21] M. Fukumoto, E. Nishioka, T. Matsubara, Surf. Coat. Technol. 120 /121 (1999) 131. [22] S. Sampath, X. Jiang, Mater. Sci. Eng. A 304 /306 (2001) 144. [23] P. Fauchais, A. Vardelle, B. Dussoubs, J. Therm. Spray Technol. 10 (1) (2001) 44. [24] Mo /O Phase diagram, Bull. Alloy Phase Diagrams 1 (1980) 85. [25] R. Goswami, H. Herman, S. Sampath, X.Y. Jiang, Y. Tian, H. Halada, Surf. Coat. Technol. 141 (2001) 220. [26] X. Jiang, Y. Wan, S. Sampath, H. Herman, Thin Solid Films 385 (1 /2) (2001) 132.
Development of process maps for plasma spray: case study for molybdenum S. Sampath a,*, X. Jiang a,1, A. Kulkarni a, J. Matejicek a,2, D.L. Gilmore b,3, R.A. Neiser b a
Department of Materials Science and Engineering, Center for Thermal Spray Research, State University of New York, Stony Brook, NY 11794-2275, USA b Thermal Spray Research Laboratory, Sandia National Laboratory, Albuquerque, NM, USA Received 11 April 2002; received in revised form 23 August 2002
Abstract A schematic representation referred to as ‘‘process maps’’ examines the role of process variables on the properties of plasmasprayed coatings. Process maps have been developed for air plasma spraying of molybdenum. Experimental work was done to investigate the importance of such spray parameters as gun current, primary gas flow, auxiliary gas flow, and powder carrier gas flow. In-flight particle temperatures and velocities were measured and diameters estimated in various areas of the spray plume. Empirical models were developed relating the input parameters to the in-flight particle characteristics. Molybdenum splats and coatings were produced at three distinct process conditions identified from the first-order process map experiments. In addition, substrate surface temperature during deposition was treated as a variable. Within the tested range, modulus, hardness and thermal conductivity increases with particle velocity, while oxygen content and porosity decreases. Increasing substrate deposition temperature resulted in dramatic improvement in coating thermal conductivity and modulus, while simultaneously increasing coating oxide content. Indentation reveals improved fracture resistance for the coatings prepared at higher substrate temperature. Residual stress was significantly affected by substrate temperature, although not to a great extent by particle conditions within the investigated parameter range. Coatings prepared at high substrate temperature with high-energy particles suffered considerably less damage in a wear test. The mechanisms behind these changes are discussed within the context relational maps, which have been proposed. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Process maps; Plasma spray; Thermal spray; Molybdenum; Residual stress
1. Introduction Thermal spraying is a well-established technology for the production of overlay protective coatings and is used extensively in industry servicing a range of components. Metallic and ceramic coatings allow engineering components to operate under extreme conditions of wear, corrosion and high-temperature exposure. More recently, sprayed coatings and multilayers are being considered for functional surfaces as well.
* Corresponding author 1 Presently at Caterpillar, Inc., Peoria, IL, USA. 2 Presently at Institute of Plasma Physics, Prague, Czech Republic. 3 Presently at CNA Corp., Alexandria, VA 22302, USA.
The use of thermal-sprayed coatings to date has been based on the concept of life extension of the engineering component, i.e. sprayed coatings are not considered as part of the component design. New demands arising particularly from the automotive and aerospace industries now require that the sprayed coatings are ‘‘prime reliant ’’ and that they be included during the design of the system. This requires considerable enhancement of the reliability and reproducibility of the coatings, as well as establishment of a knowledge base on their intrinsic materials properties and behavior. This has stimulated considerable research in the field. Thermal spraying is a highly complex deposition process with a large number of interrelated variables. Traditionally, satisfactory deposit quality is achieved
0921-5093/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 0 2 ) 0 0 6 4 2 - 1
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
through a design of experiment (DoE) approach. Numerous engineering parameters, such as power level, primary plasma gas type and flow rate, auxiliary plasma gas type and flow rate, stand-off distance, etc. have to be tried empirically and systematically through parameter matrix to determine the optimal parameters. The insight of the intermediate subprocesses and the corresponding mechanisms are largely unknown. The requirement of prime reliant coatings has created a need to develop strong scientific correlations among these complex parameters. This requires a concerted, integrated interdisciplinary approach, and this study demonstrates such an approach for a specific spray system and material. Process maps are a vehicle to establish these correlations based on sound scientific principles. From a scientific point of view, particle velocity (Vp) and temperature (Tp), substrate temperature (Ts) are the most important parameters, which determine the deposit build-up process and deposit properties. Particle temperature and velocity affect the deposit efficiency as well as the microstructure. Particle parameters affect almost every single subprocess in the deposition process, such as droplet spreading/solidification, droplet/substrate or previously deposited layer interaction, while substrate temperature has been found to have significant influence on splat morphology, deposit microstructure and properties [1 /3]. With an increase of substrate temperature, splat morphology changes from highly fragmented to contiguous, disk-like shape [1 /3], leading to enhanced deposit integrity, microstructure and properties [3,4].
55
ment, a good process map can enhance stability and reliability while also allowing process modifications to be successfully performed with a minimum of downtime, leading to improved economics. In a research environment, process maps allow one to validate theoretical models. The combination of experimental and theoretical data may allow interpolation and extrapolation to untried conditions. The knowledge gained in the development of a process map can be both fundamental and practical in nature. In this collaborative study, an effort was made to determine the relationships between spray parameters, in-flight particle characteristics, deposit microstructures and properties in a systematic manner. The aim of the study is to develop a more fundamental understanding of plasma /particle interactions, droplet /substrate interactions, deposit formation dynamics and microstructural development, as well as final deposit properties. The focus of this first part of this paper is the link between the processing parameters and the spray plume. The results of this part yield what is referred to as a firstorder process map for plasma spray. Based on these results, a set of process conditions was identified to generate splats and coatings to establish the linkages between process microstructures and properties. The correlations resulting from the second set of results yield a second-order process map . Fig. 1 illustrates the philosophy of such process maps for plasma spray. Ultimately, the combination of such maps provided an integrated set of understanding of the complex interactions in thermal spray and allow both fundamental understanding of the process as well as to provide tools for their utilization in the manufacturing sector.
1.1. Process maps A process map is an integrated set of relationships that link processing to properties and ultimately to performance. The map can consist of both experimental and computational data. Several types of linking relationships comprise a thermal spray process map. First, how do process conditions affect the particle plume? Experimental studies of this kind have become more common as particle diagnostics techniques have improved [5 /8]. Also, advances in the power of computational models have led to more theoretical studies of plasma/particle interactions [9,10]. Next, how does the particle plume interact with the substrate to produce the coating microstructure? And last, how does the microstructure behave when tested (i.e. properties and performance) and why? Studies linking particle characteristics to splat formation and coating microstructure and properties have also become more common in recent years [6,7,11 /13]. Process maps are useful because they allow one to more intelligently control the process in a non-research environment. For example, in a production environ-
2. Experimental 2.1. Torch set-up Plasma spraying was performed with a Praxair TAFA (formerly Miller Thermal) SG-100 plasma torch in a vertical orientation at Sandia National Laboratory. The Albuquerque, NM ambient atmospheric pressure for spraying was 83 kPa (12.2 psi). The torch hardware consisted of a 730 anode, a 720 cathode, and a 112 (straight) gas injector ring. The main plasma gas and the powder carrier gas were argon, the auxiliary plasma gas was helium, and the powder carrier gas was argon. Gas flow rates were measured using pressure transducers and jeweled critical orifices calibrated using a primary standard at Sandia Labs. The helium and argon were technical grade. A new anode/cathode set was used, and it was ‘‘broken in’’ prior to data collection by running for 30 min at 500 A, argon only. A straight powder feed tube with a 3.0 mm inner diameter was used, and powder injection was at 188
56
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
Fig. 1. Schematic overview of process maps for plasma spray.
forward of normal through a 1.8 mm diameter port. A plasma-densified fine size cut molybdenum powder (Osram Sylvania, #SD152) was used for the work presented in this paper. The volumetrically weighted powder size distribution as measured by a laser diffraction system (Model LS-100, Coulter Corp., Miami, FL) is given in Table 1. 2.2. Particle diagnostics Temperature, velocity, and diameter were measured for individual particles in the air plasma spray plume using a DPV-2000 diagnostics system (Tecnar Automation Ltd., St Hubert, Que.). The velocity is determined by time-of-flight between mask slits. The temperature is determined by two-color pyrometry. The diameter is calculated by integration of a pyrometry peak to give the radiated energy for a given particle. This is then used in conjunction with the temperature measurement to give the particle size. The depth of field and the diameter of the measurement region are both in the tenths of millimeter range. Further details of this system are available elsewhere [14]. The coefficient for the diameter calculations was calibrated by spraying with a different molybdenum powder which had a very narrow size cut. Fig. 2 compares the size distribution of the SD152 molybdenum powder before spraying with a typical size distribution for the in-flight particles.
Fig. 2. Molybdenum powder size distributions.
was fixed at 50 standard liters per minute (slpm) of argon and the powder feed rate was fixed at 0.18 g s 1 (1.5 lbs h1). A uniform-precision central-composite design was used to construct a 44-point experimental matrix in this three-dimensional parameter space, including an ‘‘outer’’ three-level full factorial cube (27 points), an ‘‘inner’’ two-level full factorial cube (eight points), and nine more repetitions of the central point (700 A, 18 slpm He, 2 slpm powder carrier gas) in order to distinguish the underlying measurement error from lack of fit in later empirical modeling. The parameter values at which particle measurements were collected are given in Table 2.
2.3. Design of experiments 2.4. Particle measurements The plasma torch parameters of importance to be investigated were torch current, auxiliary helium flow, and powder carrier gas flow (argon). During the experiments to determine the effects of these parameters on the particle characteristics, the main plasma gas flow
Having determined the parameter values at which particle measurements were to be made, the order in which measurements were randomized. The stand-off distance for the diagnostics sensors was fixed at 100 mm
Table 1 Molybdenum powder size distribution Volumetric mean (mm)
10% (mm)
25% (mm)
50% (mm)
75% (mm)
90% (mm)
2899
40
35
28
22
16
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66 Table 2 Torch parameters for DoEs particle measurements Parameter
Outer cube Inner cube
Current (A)
Helium (slpm)
Carrier gas (slpm)
500, 700, 900 580, 820
12, 18, 24 14.4, 21.6
1.0, 2.0, 3.0 1.4, 2.6
(4.0 in.) downstream of the exit plane of the torch. For each set of torch parameters, the sensors were then ‘‘auto-centered’’. That is, the sensors were positioned by an automated routine, which centered the measurement volume in the region of maximum particle flow. A map of particle flow in a planar cross-section of the plume is shown in Fig. 3. Note that in this figure, the central axis of the torch would pass through (0,0), and the powder injection tube would be parallel to the Y -axis at X /0. Thus, the measurements for this particular run would have been made at a position 6 mm (14 in:) from the central axis of the torch, in-line with the powder injection tube. 2.5. Fabrication of splats and deposits Based on the results of the first-order process maps, three conditions were selected with varying particle energies. These are nominally designated as low, medium and high. Splats were also collected on substrates nominally at three substrate temperatures */low: 115 8C, medium: 325 8C, high 465 8C. Details of the plasma and spraying condition are listed in Table 3. Characterization includes the observation and measurement of splats and the corresponding craters formed on the substrates. The procedure is as follows. First identification marks were made on the substrates by indentation so that the splats can be numbered and
Fig. 3. Distribution of particle flow rates within a plume cross-section at 100 mm stand-off. Data collected at ‘‘central’’ conditions (700 A, 18 slpm He, 2.0 slpm carrier gas).
57
located, subsequently splats were observed with optical microscopy and scanning electron microscopy (SEM) and dimensions were measured. Following this, the splats were etched away with saturated nitric acid and craters were characterized. The dimensions of splats and craters were measured with Zygo New View 200 noncontact surface profiler (a scanning white-light interferometer) [15]. Hardness measurement was carried out with a Buehler Micromet II microhardness tester using Vickers indentor and a 500 g load on the cross-section of the coating. High load indentation was carried out with a Mitutoyo AVK-C2 hardness tester at 20 kg load. Thermal conductivity measurements of the coatings were carried out on freestanding specimens by a laser flash technique, using Holometrix Thermal Properties Instrument. Open porosity was measured by mercury intrusion technique, using the Quantachrome Autoscan 33 porosimeter. Residual stresses were measured in 0.1 mm thick coatings on stainless steel. X-ray diffraction technique ‘‘sin2 c ’’ [16] was applied, using Siemens D500 diffractometer with Ni-filtered Cu radiation, reflection from (3 2 1) crystal planes with elastic constants E /313 GPa and n/0.31 and 10 sample tilts from c //528 to 528.
3. Results The first part of this paper examines the relationships between torch operating conditions and particle conditions as assessed by on-line in-flight diagnostics. As described in Section 2, a detailed evaluation of the particle conditions at the 44 DoEs was conducted. Based on the analysis of these results, a first-order process map relating particle velocity and temperature was generated and this served as the basis to produce splats and deposits for the second-order process map experiments. These results are summarized below. Fig. 4 displays particle temperature contours in a plume cross-section sprayed at the ‘‘central’’ conditions of the DoE. Comparing with Fig. 3, it can be seen that the region of hottest particles is shifted only slightly away from the region of maximum flow, and that all of the particles in this central area of the plume appear to be above the melting temperature of molybdenum (2610 8C). A quantitative estimate of the significance, or lack thereof, of particle segregation in the plume has been made by calculating the flow-weighted averages of particle temperature, velocity, and diameter at 121 points throughout the plume (a 10/10 mm2 with 1 mm point spacing). Table 4 shows that the standard deviations of the particle data are no more than 8% of the mean values. That is, although particle segregation exists, it is small, and the largest deviations in particle
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
58
Table 3 Plasma spraying parameters for splats and deposits Plasma power/parti- Current cle energy (A)
Plasma gas and flow rate
Carrier gas and flow Average particle temrate (lpm) perature (8C)
Average particle velo- Substrate temperacity (m s 1) ture (8C)
Low
540
Ar: 3.0
2780
130
115, 325a
Medium
700
Ar: 2.0
2980
153
115a, 325a
High
860
Ar: 50, He: 12 lpm Ar: 50, He: 18 lpm Ar: 50, He: 26 lpm
Ar: 1.0
3120
183
115a, 325a, 465
lpm: liter per minute. a Coatings were made at these Ts conditions.
Fig. 4. Distribution of particle temperatures within a plume crosssection at 100 mm stand-off and ‘‘central’’ conditions. Fig. 5. Mean particle temperatures and velocities at the various spray conditions are highly correlated. Table 4 Flow-weighted particle statistics for Figs. 3 and 4
Temperature Velocity Diameter
Mean9S.D.
S.D./mean (%)
2942969 8C 154911 m s 1 3393 mm
2.3 7.3 8.1
temperature and velocity occur at the edges of the plume where there are few particles present. 3.1. Correlations Fig. 5 indicates that for the 35 different spray conditions at which particle data were collected, the mean particle temperatures and velocities were highly correlated, as might be expected. This makes it difficult to separate the effects of particle kinetic energy and thermal energy on coating microstructure and properties. The mean particle temperature was found to be above the melting point of molybdenum for all spray conditions. Fig. 6 shows that, for a given run, particle velocity and diameter were also found to be correlated. The data fit fairly well to a curve derived from simple aerodynamic drag, with velocity being proportional to the inverse square root of the diameter. There is a signifi-
Fig. 6. Particle velocities for a given run agree with a simple aerodynamic drag model (V 8/1/d1/2). Data collected at ‘‘central’’ conditions (700 A, 18 slpm He, 2.0 slpm carrier gas) at spray pattern center (maximum particle flux).
cant amount of scatter in the data about this curve, presumably due to the statistical distribution of particle injection trajectories. 3.2. Empirical model The data from Fig. 5 were used to create an empirical model of the effects of spray parameters on particle
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
59
temperature and velocity (little effect on particle diameter was observed). The parameters (current, helium flow, powder carrier gas flow) were first normalized for the ranges investigated so that the relative importance of each term in the model could be accurately gauged. The particle characteristics were then fit to a quadratic surface in the three-dimensional parameter space: Tp =Vp C0 C1currentC2helium C3carrier gasC4currentcurrent Fig. 8. Pareto plot of parameter effects on particle velocity.
C5currenthelium C6currentcarrier gas C7heliumhelium C8heliumcarrier gas C9carrier gascarrier gas;
(1)
where all parameters are normalized and the coefficients are as given in Figs. 7 and 8. For example, referring to Eq. (1) and Fig. 7, for spray conditions of 800 A, 21 slpm helium, and 2.5 slpm powder carrier gas, the model would predict a particle temperature of 2999 8C. All the parameters in this example would have normalized values of 0.50 (e.g. 800 A is half-way between the ‘‘central’’ condition of 700 A and the highest current used in the DoE, 900 A). 3.3. Relative importance of parameters Figs. 7 and 8 show that the plasma current is the dominant influence on the particle temperature and velocity, as might be expected. In the range of values, which were studied, increasing powder carrier gas flow decreased both temperature and velocity, indicating that this system is fairly sensitive to the trajectory of the injected particles. The secondary gases impart good heat transfer properties thereby increasing plasma voltage and energy. However, helium flow has a surprisingly small effect on particle temperature*/an increase of 6 slpm only results in a 9 8C rise. Fig. 9 plots the predicted values for particle temperature and velocity from the empirical model given above. Values were calculated over the full range of parameter space investigated: 500/900 A, 12 /24 slpm helium, 1.0 /
Fig. 9. Predicted and measured values for mean particle temperature and velocity.
3.0 slpm carrier gas. For comparison, the 44 experimentally measured data points are also shown. As noted previously, temperature and velocity are highly correlated. However, the predicted values show that by properly adjusting the spray parameters, one may vary temperature and velocity independently within small ranges, more so at low values than at high values. 3.4. Test and use of model In order to test the model, new experimental data were taken at previously untested points in parameter space. Some of these new points were within the limits of the original DoE cube (testing interpolation), and some lay outside those limits (testing extrapolation). The new measured values were found to agree with predicted values within 2% on average for temperature, and within 3% on average for velocity. In general, interpolated predictions were found to be more accurate than extrapolated ones, as might be expected. 3.5. Fabrication of splats and deposits
Fig. 7. Pareto plot of parameter effects on particle temperature.
Once process maps linking the torch parameters to the particle characteristics were completed, conditions were chosen for the synthesis of samples. These conditions are indicated in Table 3 and illustrated in Fig. 10.
60
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
Fig. 10. Processing condition for process map samples. Particle energies are 350 J mol 1 (low), 540 J mol1 (medium) and 700 J mol 1 (high) and substrate temperatures vary between 115 8C (low) and 325 8C (high).
As in the DoE work, the torch axis was vertical, and the torch /substrate stand-off distance was 100 mm (4.0 in.). Substrates were mounted eight at a time in a carouseltype fixture, roughly 150 mm (6 in.) in diameter and rotating about a horizontal axis. Thus, a given substrate was normal to the torch axis every one-eighth of a revolution. As substrate temperature has been found to have a large influence on coating microstructures and properties [17,18], the surface temperatures (Ts) of the samples were monitored and controlled by an air-cooling feedback system. This consisted of an infrared pyrometer which measured the temperature of a given substrate (or top surface of the deposit) at a position one-quarter of a revolution after it had passed under the spray plume. This temperature data were then used to control the pressure of the cooling air jet, which was positioned 1808 opposite to the torch. With this system, it was typically possible to hold a desired temperature within 5 8C. However, the range of substrate temperatures achievable were limited in some cases by the available cooling air flow and the particular heat input for a given set of plasma conditions. For a given set of conditions, three types of deposit samples were sprayed: splats (to study the effect of inflight particle characteristics and substrate temperature on morphology), thin coatings (for wear and residual stress measurements), and thick coatings (for metallography, porosity and thermal conductivity measurements). The specimen configurations are schematically illustrated in Fig. 10. Splats were produced by passing the torch once over the substrate carousel (rotating at 240 rpm) at a traverse rate of 100 mm s 1 (4.0 in. s 1) and a powder feed rate of approximately 0.05 g s 1 (0.4 lbs h1). Thin coatings, nominally 100 mm (0.004 in.) thick, were produced by passing the torch over the substrates (150 rpm) about 50 times at a traverse rate of 25 mm s 1 (1.0 in. s 1) and a powder feed rate of approximately 0.18 g s 1 (1.5 lbs h1). Thick coatings,
nominally 1.3 mm (0.05 in.) thick, were produced with about 250 passes at 7.5 mm s 1 (0.3 in. s1), a feed rate of 0.18 g s 1 (1.5 lbs h 1), and a carousel rotational speed of 180 rpm. Particle temperature, velocity, and diameter were measured for every spray run to insure that they were within the target ranges. 3.6. Droplet/substrate interaction and the establishment of adhesion It was found that molybdenum splat morphology is sensitive to particle conditions and substrate temperature. At low substrate temperature (Ts), all splats show highly splashed morphology with a fragmented core structure and debris scattered radially. With an increase in substrate temperatures and decrease of particle energy, contiguous splat is more readily formed as schematically shown in Fig. 11.
Fig. 11. Process map illustrating splat morphologies at various processing conditions. Particle energies are 350 J mol 1 (low), 540 J mol 1 (medium) and 700 J mol1 (high) and substrate temperatures vary between 115 8C (low) and 325 8C (high).
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
It is found that substrate is significantly melted and deformed upon the impact of droplet, which leads to the formation of flower-like splats and craters. On average, only about 36 /53% of the areas covered by splats were in good metallurgical/mechanical contact with substrate, as indicated by substrate alloying and/or deformation, with higher particle energy/higher Ts combination shows higher effective contact area ratio. Crater volume can reach up to 35% of the impacting droplet volume, indicating very severe substrate deformation [19].
3.7. Coating microstructure and integrity Fig. 12 shows the cross-section of coatings prepared at Ts /115 8C at three particle energy conditions. All of them display lamellar microstructure. With the increase of particle energy, the lamellae in the coatings become slightly thinner due to the higher particle velocity. The number of interfaces is larger for higher particle energy for a given thickness. It appears that porosity is slightly higher in the low particle energy coating. Inter-lamellar pores can be discerned clearly in all three coatings. Coatings produced at higher Ts show similar microstructure. Indentation method was used to reveal possible differences in coating mechanical response originated from different microstructure at various PE and Ts conditions. The impressions and surrounding coating structures are shown in Fig. 13. At the load of 500 g, numerous inter-lamellar cracks can be found around the impression for all the coatings produced at Ts /115 8C regardless of the particle energy level. Meanwhile, for the coatings produced at higher Ts, no cracks were observed around the impression. It is noticed that interlamellae boundaries are more readily seen in the higher Ts coatings. This feature is probably due to the higher oxide level on the interface associated with higher Ts coatings (see discussion in the oxidation section). With the increase of load to up to 20 kg, low Ts coatings show similar cracking pattern and higher Ts coatings show long cracking initiated from the tips of indentation diagonal and propagating parallel to the deposit/sub-
61
strate interface along lamellae/lamellae boundaries as shown in Fig. 14 (medium PE). Initiation of cracking is determined by the local stress level and the coating fracture toughness. The existence of many short cracks around the low Ts coatings indentations indicates evidently that the adhesion between lamellae and lamellae is much weaker compared with coatings produced at higher Ts. The stress is relaxed even when it is small and therefore no large stress built up near the indentation tip. In the case of higher Ts coating, stress is built-up and cracks initiated at the indentation tip where maximum stress exists. This observation is consistent with our observation before in another study, where, high Ts molybdenum coating showed trans-lamellae fracture feature as compared with inter-lamellae mode in the low Ts deposit [3]. The fact that cracks are parallel to the deposit/substrate interface is a manifestation of the anisotropy in deposit structure and property. 3.8. Residual stress The effects of deposition temperature and particle energy on residual stress are presented in Table 5 and Fig. 15. As can be seen from Fig. 15a, deposition temperature has quite a dramatic effect on the coating residual stress. When the temperature increases from 120 to 280 8C, the stress shifts from tensile to compressive and becomes even more compressive above 300 8C. This trend can be explained by increasing contribution of thermal stress, which is proportional to the difference in coefficients of thermal expansion (CTE) of the substrate and the coating, temperature drop during post-deposition cooling and coating modulus. Since the coating’s CTE is lower than the substrate’s, the thermal stress is compressive and its magnitude increases with deposition temperature due to larger temperature drop after the deposition. Increasing modulus with deposition temperature [3] can explain the increase in slope in Fig. 15a. The effect of particle energy is rather negligible, as shown on Fig. 15b. Only a small trend of decreasing stress with particle energy was observed at lower deposition temperature; however,
Fig. 12. Cross-section of coating made at three particle energy conditions at Ts /115 8C etchant: Murakami’s. (a) low PE (350 J mol 1), (b) medium PE (540 J mol1), and (c) high PE (700 J mol 1).
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
62
Fig. 13. Indentation mark and cracking pattern on coatings produced at different particle energies (PE) and Ts conditions, load 500 g. (a) Low PE (350 J mol 1), (b) medium PE (540 J mol1), and (c) high PE (700 J mol 1).
Table 5 Residual stresses for coatings produced at various spray conditions Particle energy
Ts (8C)
Low
120 275
13
20
Medium
120 285
12 92
16 15
High
120 275 330
10 92 331
19 18 18
Stress (MPa)
S.D.
the differences were smaller than experimental error. The particle energy effects may become more significant in thick coatings, where the associated changes in splat/ coating formation will affect the microstructure and thus mechanical properties. Compressive stress in the coating may be beneficial for its mechanical properties, since it would close the existing cracks and suppress crack propagation, but too
high magnitude can lead to delamination at the coating/ substrate interface. The stress state and mechanical properties can be easily controlled by the deposition temperature and therefore can be tailored to specific application. 3.9. Oxygen content Oxygen content of coatings produced at various conditions is plotted in Fig. 16. It is clear that at 115 8C, with an increase of particle velocity, oxygen content decreases. For the same particle velocity, coating produced at higher Ts show higher oxygen content. This result corroborates that both in-flight particle oxidation and post-depositing surface oxidation contribute to the overall coating oxidation. It is shown through modeling that in-flight oxidation is primarily controlled by the residence time of molten particle in the oxygen containing flame [20]. Since the nozzle/substrate distance is fixed in all the conditions, high velocity particles will have low oxygen content. At
Fig. 14. Indentation mark and crack pattern on coatings produced at two Ts conditions, particle energy level: medium (540 J mol 1); load: 20 kg.
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
63
Fig. 17. Microhardness of coatings produced on various particle energy and substrate conditions.
Fig. 15. (a) Residual stress vs. deposition temperature for different particle energies. (b) Residual stress vs. particle energy for different deposition temperatures. Fig. 18. Mercury intrusion porosity of coatings produced on various particle energy and substrate conditions.
3.10. Hardness and porosity As shown in Figs. 17 and 18, with the increase of particle kinetic energy (calculated from particle velocity and size data), coating porosity decreases significantly and coating hardness increases slightly for the two Ts investigated. Substrate temperature shows a dramatic effect on coating hardness. With an increase of Ts from 115 to 325 8C, hardness increases by more than 20%. But the porosity change is not as dramatic. Fig. 16. Oxygen content of coatings produced on various particle energy and substrate conditions.
3.11. Thermal conductivity Table 6 lists the thermal conductivity data for coatings made at various conditions. These values are only
higher Ts, post-depositing surface oxidation is much more significant, and so the overall oxygen content is increased. The reason for the higher oxygen content at 180 m s1 coating as compared with 160 m s 1 coating at high Ts is most likely associated with marginally increased Ts for 183 m s 1 coating (325 8C) compared with that for the 153 m s 1 coating (310 8C).
Table 6 Thermal conductivity of the coatings Thermal conductivity (W m 1 K 1)
Ts 115 8C
Ts 325 8C
Medium PE High PE
13.0 16.6
46.5 45.3
64
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
about 10/30% of that for sintered molybdenum, associated with the existence of many inter-lamellae interfaces and pores. It appears that thermal conductivity is not significantly affected by particle energy but very sensitive to Ts. Thermal conductivity tripled with the increase of substrate temperature from 115 to 325 8C. Thermal conductivity is an indication of inter-lamellae contact quality. The dramatic increase of thermal conductivity reveals significant enhancement of interlamellae bonding and adhesion.
4. Discussion 4.1. Effect of particle conditions and substrate temperature
reveal detectable amount of MoO3 inside the coatings [25]. Perhaps the low melting point oxide melts and dissolves into the impinging droplet, assisting to form good metallurgical bonding between splats and improve thermal conductivity. If the oxide exists as a solid film at the interface, the coating thermal conductivity would have been reduced. Similar coating property enhancement at higher Ts was reported in partially stabilized zirconia deposits [4]. This suggests mechanism (3) is unlikely to be the only mechanism. In reality, perhaps all three mechanisms are active. However, the most plausible explanation is based on #1 the role of adsorbates and condensates. Recent works from some of the authors have shown this to be a rather dramatic effect [26]. 4.2. Synthesis second-order process maps
The results of the first-order process maps indicate that for a given plasma spray nozzle, the particle velocity and particle temperature are highly correlated. Consequently, parameters that cause an increase in one of the variable generally results in the increase in the second variable. In this work, the coupled variable of temperature and velocity is treated as particle energy and the combined effect of this parameter is addressed through integration into the second-order process maps. The particle energy has the effect of enhancing the density of the deposit and consequently an increase in elastic modulus. For the explored range of particle energy space, there is almost no effect on residual stresses. The particle oxidation is reduced in spite of increase in particle energy, given the reduced dwell time of the particles. It is evident that substrate temperature has a far more influencing effect on coating microstructure, hardness, residual stress and inter-lamellae contact than particle energy within the investigated parameter ranges. Coatings produced at higher Ts also show lower coefficient of friction and higher scratch resistance [9]. Although, increasing particle energies show effect in reducing coating porosity and increasing hardness, the improvement is not dramatic. The mechanism of improvement in inter-lamellae adhesion at higher Ts has been a subject of considerable analysis in recent years [21 /23]. Our current understanding suggests that there are three possibilities: (1) reduced adsorbates/condensates at higher substrate temperatures; (2) less trapped air in the pores due to lower air density at higher temperature, lower air density results in less severe disturbance to droplet spreading may lead to better contact; and (3) thicker splat surface oxide layer (presumably MoO3 as suggested in literature [24]) as evidenced by the higher oxygen content may help to improve the contact quality. XPS study shows trace of MoO3 on the as-sprayed coating surface, but Xray and electron diffraction of similar coatings did not
The second-order process map links the effect of key process variables on the microstructure and properties of the sprayed materials. The goal here is to identify the sensitivities of these variables on the properties of the deposit while simultaneously generating regime maps which will not only provide a fundamental understanding of intrinsic deposit properties but also enable their incorporation into design. It is envisioned that such maps will provide designers means to consider the coating into the system and also allow manipulation of the properties to achieve specific goals. Fig. 19 provides a schematic representation of one such process map. In this diagram, three attributes of the deposit are identified with respect to the processing conditions and based on their sensitivities. The lines with arrows depict a trend in the property within the process operational regime and slope suggests the sensitivity of the parameter. The dotted line represents a zero stress line where the tensile residual stress generated during the quenching of the splat is matched by the thermal mismatch stress for the steel /Mo system
Fig. 19. Illustrating 2nd-order process map for plasma spraying of molybdenum on steel substrates.
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66
at the deposition temperature. In addition, regimes of lower oxide content in the coating are also depicted in the map. The maps provide important information about the process and the resulting deposit. The key ones are identified as follows: . Increasing particle energy is a major driver for reducing porosity in the material. Increasing deposition temperature can also reduce porosity but this effect is less sensitive to that of particle energy. The elastic modulus of the deposit correlates with that of the overall porosity of the coating. . Increasing particle energy reduces the dwell time of the particles within the plasma and the associated entrainment field and reduces the in-flight oxidative processes. However, increasing deposition temperature increases the oxide content in the deposit, arising from surface oxidative phenomena. . Increasing particle energy has no effect on residual stresses in the deposit consistent with the quenching stress based residual stress in the system. At higher deposition temperatures, the thermal stress field is superimposed on the quenching stress. This can lead to either tensile or compressive overall residual stress depending on the mismatch in the substrate-coating system. . Through thickness thermal conductivity of the deposit is more strongly affected by the substrate temperature rather than an increase in overall porosity. This is due to the improvement in wetting and associated contact between lamellae within the deposit.
5. Conclusions Thermal spraying is often very sensitive to process details and substrate condition and this has a profound effect on the properties and ultimately the performance of the sprayed coating. Process maps offer a means to relate process conditions to the critical properties of the sprayed coatings in a systematic and intelligent manner. Diagrams relating torch parameters to the thermal and kinetic energy of sprayed particles can be constructed rapidly, typically in a few days to a week. These diagrams can be used to intelligently screen for processing conditions that are likely to produce a wide range of coating microstructures. This screening reduces the need for time-consuming characterization of numerous samples, frequently requiring weeks to months. In this paper, air plasma spray of molybdenum on steel substrate was studied. An empirical model was developed from experimental data to link the torchinput parameters to the properties of the spray plume. The mean velocity and temperature of the in-flight
65
particles were found to be strongly correlated. Applied current was found to have the dominant influence on particle characteristics. The trends observed from these experiments can be constructed in the form of process maps, which illustrate the correlations and identify the sensitivities of the parameters to the output variables. Linking these process variables to the deposit formation and dynamics and ultimately the deposit properties allows an integrated understanding of the process to properties. It was found that there is a strong mechanical/thermal interaction between droplet and substrate, which builds up the coating/substrate adhesion. Hardness, thermal conductivity increase, oxygen content and porosity decreases with increase of particle velocity. Increasing deposition temperature resulted in dramatic improvement in coating thermal conductivity and hardness as well as increase in coating oxygen content. The residual stress in the coatings changes from tensile to compressive as the deposition temperature increases for the steel /molybdenum system, which makes this parameter an effective means of controlling the stress level. In the investigated parameter range, substrate temperature has a larger effect on coating property as compared with particle energy. A type of process map has been generated based on this systematic study. Although this has been demonstrated for a given material /substrate combination, these concepts can be generalized. This provides a means enhance fundamental understanding of the complex interactions during thermal spray processing and illustrates sensitivities of certain process variables to critical deposit properties. The framework also allows visualization of process regimes to achieve desired coating properties. Such strategies will allow more expansive and intelligent utilization of thermal spray coatings.
Acknowledgements This research was supported by the MRSEC program of the National Science Foundation under award DMR 9632570 and DMR 00890021. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under Contract DE-AC04-94AL85000. The authors would also like to acknowledge the role of David E. Beatty in the collection of particle data and the production of test samples.
References [1] L. Bianchi, F. Blein, P. Lucchese, M. Vardelle, A. Vardelle, P. Fauchais, in: C. Berndt, S. Sampath (Eds.), Thermal Spray
66
[2]
[3] [4] [5]
[6]
[7]
[8]
[9] [10] [11]
[12]
S. Sampath et al. / Materials Science and Engineering A348 (2003) 54 /66 Industrial Applications, ASM International, Metals Park, OH, 1994, p. 569. M. Fukumoto, S. Katoh, I. Okane, in: A. Ohmori (Ed.), Proceedings of the Integrated Thermal Spray Conference, Kobe, Japan, Japan High Temperature Society, 1995, pp. 353 /358. X. Jiang, J. Matejicek, S. Sampath, Mater. Sci. Eng. A 272 (1999) 189. S. Sampath, X.Y. Jiang, J. Matejicek, Mater. Sci. Eng. A 272 (1999) 194. J.R. Fincke, W.D. Swank, D.C. Haggard, T.M. Demeny, S.M. Pandit, A.R. Kashani, in: C.C Berndt, S. Sampath (Eds.), Advances in Thermal Spray Science and Technology, ASM International, Materials Park, OH, 1995, pp. 117 /122. R.N. Wright, J.R. Fincke, W.D. Swank, D.C. Haggard, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 511 /516. M. Prystay, P. Gougeon, C. Moreau, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 517 /523. W.D. Swank, J.R. Fincke, D.C. Haggard, S. Sampath, W. Smith, in: C.C. Berndt (Ed.), Thermal Spray: A United Forum for Scientific and Technological Advances, ASM International, Materials Park, OH, 1997, pp. 451 /458. J.D. Ramshaw, C.H. Chang, Plasma Chem. Plasma Process. 12 (1993) 299. Y.P. Wan, V. Prasad, G.X. Wang, S. Sampath, J.R. Fincke, Int. J. Heat Mass Transfer 121 (1999) 559. A.C. Le´ger, M. Vardelle, A. Vardelle, P. Fauchais, S. Sampath, C.C. Berndt, H. Herman, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 623 /628. S. Sampath, J. Matejicek, C.C. Berndt, H. Herman, A.C. Le´ger, M. Vardelle, A. Vardelle, P. Fauchais, in: C.C. Berndt (Ed.), Thermal Spray: Practical Solutions for Engineering Problems, ASM International, Materials Park, OH, 1996, pp. 629 /636.
[13] M. Vardelle, P. Fauchais, A. Vardelle, A.C. Le´ger, in: C.C. Berndt (Ed.), Thermal Spray: A United Forum for Scientific and Technological Advances, ASM International, Materials Park, OH, 1997, pp. 535 /542. [14] C. Moreau, P. Gougeon, M. Lamontagne, V. Lacasse, G. Vaudreuil, P. Cielo, in: C.C. Berndt, S. Sampath (Eds.), Thermal Spray Industrial Applications, ASM International, Materials Park, OH, 1994, pp. 431 /437. [15] P. De Groot, J. Mod. Opt. 42 (1995) 389. [16] J. Matejicek, S. Sampath, J. Dubsky, J. Therm. Spray Technol. 7 (1998) 489. [17] Haddadi, F. Nardou, A. Grimaud, P. Fauchais, in: C.C. Berndt, S. Sampath (Eds.), Advances in Thermal Spray Science and Technology, ASM International, Materials Park, OH, 1995, pp. 249 /254. [18] J. Matijicek, S. Sampath, H. Herman, in: C. Coddet (Ed.), Thermal Spray: Meeting the Challenges of the 21st Century, ASM International, Materials Park, OH, 1998, pp. 419 /424. [19] X.Y. Jiang, Y. Wan, X.Y. Wang, H. Zhang, R. Goswami, H. Herman, S. Sampath, in: C.C. Berndt (Ed.), Thermal Spray, Surface Engineering via Applied Research, ASM International, Materials Park, OH, 2000, pp. 729 /736. [20] Y. Wan, X. Jiang, H. Zhang, S. Sampath, V. Prasad, J. Fincke, Metall. Mater. Trans. B 32 (2001) 475. [21] M. Fukumoto, E. Nishioka, T. Matsubara, Surf. Coat. Technol. 120 /121 (1999) 131. [22] S. Sampath, X. Jiang, Mater. Sci. Eng. A 304 /306 (2001) 144. [23] P. Fauchais, A. Vardelle, B. Dussoubs, J. Therm. Spray Technol. 10 (1) (2001) 44. [24] Mo /O Phase diagram, Bull. Alloy Phase Diagrams 1 (1980) 85. [25] R. Goswami, H. Herman, S. Sampath, X.Y. Jiang, Y. Tian, H. Halada, Surf. Coat. Technol. 141 (2001) 220. [26] X. Jiang, Y. Wan, S. Sampath, H. Herman, Thin Solid Films 385 (1 /2) (2001) 132.