Sie sind auf Seite 1von 10

Surface and Coatings Technology 111 (1999) 6271

Kinetic spray coatings


T.H. Van Steenkiste a, J.R. Smith a,*, R.E. Teets a, J.J. Moleski a, D.W. Gorkiewicz a, R.P. Tison a,
D.R. Marantz b, K.A. Kowalsky b, W.L. Riggs, II c, P.H. Zajchowski d, B. Pilsner e,
R.C. McCune f, K.J. Barnett g
a General Motors R&D Center, Warren, MI, USA
b Flame-Spray Industries, Inc., Port Washington, NY, USA
c TubalCain Co., Loveland, OH, USA
d Pratt and Whitney Division, East Hartford, CT, USA
e GE Aircraft Engines, Cincinnati, OH, USA
f Ford Motor Co., Dearborn, MI, USA
g National Center for Manufacturing Sciences, Ann Arbor, MI, USA
Received 12 March 1998; received in revised form 14 August 1998; accepted 8 September 1998

Abstract
Coatings have been produced by entraining metal powders in an air flow which is accelerated by a de Laval type of nozzle. The
particles are not melted or thermally softened prior to impingement onto the substrate. The coating process depends primarily on
the kinetic energy of the incident powders. The coatings have low oxide content and low thermal stress, and can exhibit relatively
low porosity and high adhesion. The mechanism by which the coatings are formed is not well understood, and it is the goal of
this work to provide some insights into this mechanism. We have produced a new high-velocity spray apparatus which allows the
spray parameters to be controlled and monitored for the first time. This, together with our simulations of air and particle velocities
and temperatures, has provided new information on the coating process. Al, Cu, and Fe powders were sprayed onto Al, brass,
Cu, and steel substrates. A threshold behavior was observed for coating deposition as a function of nozzle inlet air temperature,
with a roughly linear behavior above the threshold. Results are obtained as a function of nozzle inlet air pressure and temperature,
powder feed rate, and nozzlesubstrate stand-off distance. The effect of the choice of substrate metal was relatively weak in our
experiments. Results seem consistent with necessary inelastic processes such as plastic deformation and/or partial melting of the
powder particles upon collision with the substrate. More research is needed to define the relative importance of these phenomena
or of other possible mechanisms. 1999 Published by Elsevier Science S.A. All rights reserved.
Keywords: Kinetic spray; Coatings; Cold gas; Dynamic spraying; Impact fusion; Thermal spray

1. Introduction
Coatings are being increasingly relied upon by a
variety of industries to provide thermal and electrical
insulation as well as corrosion and wear resistance [1,2].
These coatings are often applied via a thermal spray
process, whereby a powder or wire feedstock is melted
by a flame, plasma, or electric arc, and the droplets are
subsequently accelerated by a high-velocity gas. As the
droplets impinge upon a substrate, they flatten and
solidify in times of the order of a microsecond, forming
a relatively dense and adherent coating. Here we describe
a related but different process in which the powder
* Corresponding author. Fax: +1 8109863091;
e-mail: john_v._smith@notes.gmr.com

feedstock is neither melted nor softened prior to


impingement onto the substrate. The particles are
accelerated through a de Laval type of nozzle, by a
carrier gas, to speeds of around 600 m/s when the carrier
gas is air, or higher speeds when lighter carrier gases
such as He are used. The conversion of the particulate
kinetic energy to mechanical and thermal deformation
of the particles upon striking of the substrate again
leads to relatively adherent, low porosity coatings when
depositing metallic coatings. This provides coatings of
low oxide content and low thermal stress in the coatings
and substrates. Moreover, the phase of the powdered
material can be retained in the coating. As the primary
energy source is the kinetic energy of the particles, we
have named the process kinetic spray.
The first proposal in the patent literature for combin-

0257-8972/99/$ see front matter 1999 Published by Elsevier Science S.A. All rights reserved.
PII S 02 5 7 -8 9 7 2 ( 9 8 ) 0 0 70 9 - 9

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

ing thermal and kinetic energies of high-velocity particles


to make coatings that we are aware of is due to Smith
et al. [3]. Concentrating on the kinetic energy component, Rocheville [4] employed high-pressure air to make
metal and lubricant coatings. Browning [5,6 ] proposed
injecting powder into high-velocity gases heated by an
internal burner to a temperature below the melting
point, providing a sufficient velocity to achieve an impact
energy transformation to raise the temperature enough
to fuse the material into a dense coating. Alkhimov and
coworkers [7,8] proposed injecting powders into electrically heated high-pressure gases, again to temperatures
below the melting point, to make dense coatings. See
also Refs. [9,10]. The process described here could be
thought of as the next generation of the Alkhimov and
coworkers [710] process. Perhaps the most important
improvement over the original process is instrumentation for data recording and devices for feedback control
of process variables, allowing for a systematic analysis
of the process.
Such an analysis is important, because the mechanism
by which coatings are formed from solid particles colliding with a substrate is currently not well understood.
As we will see, the aspect ratio of the particles changes
substantially upon collision and, in fact, the particles
deform into conformation with the substrate and with
each other, leading to relatively low measured porosity
values in comparison with thermal spray processes
employing liquefaction of the feedstock. There are many
questions that arise about this process. For example,
one wonders if the collision leads just to plastic deformation of the particles, with perhaps some recrystallization
due to mechanical deformation. Perhaps there is also a
kinetic energy to thermal energy transition which leads
to thermal softening, or even to partial melting.
This work, while not specifically answering these
fundamental and interesting questions, begins the process of systematically analyzing the process to seek such
answers. We have recently fabricated a new high-velocity
spray apparatus which is fully instrumented and computer controlled. This allows, for the first time, feedback
control of gas temperatures, gas pressures, and other
process parameters. Thus we can now control and
analyze the process in a relatively precise fashion. In the
following, we present results for Fe, Cu, and Al coatings
made on brass and Al substrates under a variety of
process conditions in order to carry out a statistically
designed experiment (DOE ) to provide information on
how process parameters such as density and deposit
efficiency affect coating properties. All of the results are
for air as the carrier gas.

Fig. 1. Schematic of the kinetic spray nozzle.

Fig. 2. The nozzle is a de Laval type, with an entrance


cone whose diameter decreases from 7.5 mm to a 2.8 mm
throat region. Downstream of the throat region, the
nozzle has a rectangular cross-section, increasing to
2 mm by 10 mm at the exit end. These are the dimensions
appropriate for the data discussed below. The throat
diameter for the as-fabricated nozzle was 2 mm, this
feature having eroded to a larger diameter through
particle erosion. As shown in the next section, our
computations suggest that the nozzle allows air velocities
greater than 1000 m/s to be obtained, with particle
velocities dependent on particle size and the smallest
particles having velocities approaching the maximum
air velocity. Fig. 1 also shows the locations of the nozzle
inlet pressure sensor and the gas inlet temperature
thermocouple. These two quantities are computer controlled and maintained at preset values. Also controlled
in this fashion are the inlet pressure to the powder
feeder and the powder feed rate. The fifth variable
controlled in this experiment is the nozzle (gun)substrate stand-off distance, which is readily obtained by
the substrate (target) motion control as shown. The gas
temperature is varied by an in-line heater, and again
computer controlled, with temperatures achievable as
high as 650 C. A Bauer air compressor provides pressures up to 3.4 MPa (500 psi). The highest pressure in
our system, the inlet pressure to the powder feeder, is
typically limited to 2.4 MPa (350 psi), however. The
purpose of heating the gas is not to heat the particles
but rather to increase the gas velocities, as will be seen

2. Experimental apparatus
Fig. 1 shows the kinetic spray nozzle used in this
study. This is a key component of the system shown in

63

Fig. 2. Kinetic spray system layout.

64

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

in the next section. Of course, in the process the particles


are heated as well, but to temperatures well below the
melting temperatures of any of the materials sprayed in
these experiments.

3. Velocity computations
We have carried out velocity computations for the
nozzle shown in Fig. 1. The particle velocities and
temperatures were estimated from a simple one-dimensional model. Analytic equations were used to compute
the gas velocity and temperature in the nozzle from the
inlet gas conditions and the nozzle area versus length
[11]. Particle velocities in the nozzle were calculated
from the drag forces using correlations in the literature
[12].
Particle temperatures were calculated using heat
transfer correlations [13]. This simple model gives
insight into the factors that control particle velocities
and temperatures. As expected, the gas and particle
velocities scale roughly with the square root of absolute
temperature. The inlet pressure has relatively little effect
on the particle velocities. Using helium instead of air
greatly increases the velocities. While the model provides
some physical understanding of the kinetic spray process, it ignores some potentially important factors.
Boundary layers are not included, and these could
change the effective area of the nozzle, thereby altering
the gas and particle velocities. Heat transfer from the
nozzle to the gas could also be important. At heavy
particle loadings the particles could significantly slow
the flow. Finally, the model does not consider changes
in the gas and particle velocity near the substrate.
Results for air as the gas, taking the inlet air temperature to be 527 C, the inlet pressure to be 2.0 MPa
(300 psi) and the coating powder to be Cu, are shown
in Figs. 3 and 4. Although one obtains higher velocities
using He or other lighter gases as the carrier gas, we
concentrate here on results using air because the latter
would be substantially less costly in applications. The
only experimental results obtained for a gas other than
air were with He, and these are only for one Al hardness
measurement found in Table 1. Fig. 3 contains plots of
gas or particle velocity versus distance along the nozzle,
with the origin for the nozzle taken at the center of the
throat. The top curve in the plot is the gas (air) velocity
curve. The very steep portion of the gas velocity plot
coincides with the throat region of the nozzle depicted
in Fig. 1. In this case the velocity in the throat has
saturated at the speed of sound at the throat temperature, which is seen to be 680 K in Fig. 4. In Fig. 3 one
can see that the smallest Cu particles, 1 mm in diameter,
have nearly the gas velocity at each nozzle location, and
the larger the particle the lower the particle velocity.
This makes sense because the gas is accelerating the

Fig. 3. Computed air and particle velocities versus distance along the
nozzle shown in Fig. 1 as described in the text. For the particle velocity
curves, particle diameters are denoted in mm.

Fig. 4. Computed air and particle temperatures versus distance along


the nozzle shown in Fig. 1 as described in the text. Particle diameters
listed are in mm.
Table 1
Results of Vickers hardness measurements (GPa)
Average

Std.

Max.

Min.

Pure material
Aluminum
Copper
Iron

0.3
0.81
1.33

0.03
0.05
0.1

0.34
0.86
1.54

0.24
0.71
1.2

Coating
Aluminum (He)
Aluminum (air)
Copper
Iron (430 C )
Iron (650 C )

0.45
0.48
1.02
1.22
1.47

0.09
0.03
0.06
0.07
0.14

0.55
0.53
1.13
1.33
1.68

0.26
0.44
0.95
1.13
1.16

particles by drag, and the heavier particles accelerate


more slowly. These computations were validated by
measuring particle velocities for known particle diame-

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

ters. Reasonable agreement between the results of the


measurements and the computations was obtained. One
can then perhaps understand why increasing gas temperatures yield improved deposition efficiency. The gas
temperature affects the drag force on the particles by
changing the gas density and velocity and, to a small
extent, changing the drag coefficient. These factors are
included in our model. Increasing the gas temperature
always increases the particle velocity. The velocity of
sound in a gas increases as the square root of the gas
temperature and the throat velocity is typically saturated
at the velocity of sound. Moreover, one can expect the
curves of Fig. 3 to scale roughly with the throat velocity.
Thus one can expect that the particle velocities will
increase roughly as the square root of the inlet gas
temperature, as noted above. To be more specific, with
our nozzle using air at 300 psi and 800 K, we compute
that a 1% increase in gas temperature will increase the
velocity of 10 mm copper particles by 0.38%. A 1%
increase in gas temperature corresponds to a 0.50%
increase in the square root of gas temperature, so the
10 mm Cu particles velocity increases by a percentage
which is close to but somewhat less than the
percentage increase in the square root of the gas temperature. Our computations show that a 50 mm particles
velocity would increase by 0.23%.
We will see below that the coating deposition efficiency, as measured by thickness increase per pass,
increases with increasing inlet gas temperature, which is
consistent with the preceding discussion. It will be
discussed below why increased incident velocity might
be expected to increase deposition efficiency. Note that
Fig. 4 suggests that gas temperatures and temperatures
of the smallest Cu particles are below room temperature
upon exiting the nozzle. Again, all boundary effects,
such as an expected stagnation zone in front of the
substrate (target), have been ignored. In a stagnation
zone there would be a kinetic to thermal energy transition, and one would expect the gas and the particle to
heat up as a result. In fact, we have seen experimentally
that the gas heats the substrate. Note, however, that the
larger particles have temperatures approaching that of
the inlet gas.

65

(a)

(b)

4. Powder and coating properties


SEM pictures of the powders used are shown in
Fig. 5. The approximate cumulative size range distribution for the Al powder was 10% less than 9 mm, 50%
less than 20 mm, and 90% less than 40 mm. The Cu and
Fe powders were sieved to 325 mesh (45 mm).
Representative optical micrographs of kinetic sprayed
coatings are found in Fig. 6. Clearly the morphology is
quite different from that of the powders, with the aspect
ratio of the impacted particles being substantially higher

(c)
Fig. 5. SEM pictures of powders we kinetic sprayed to make the coatings discussed here. All magnifications are 500: (a) Al powders,
20 mm average diameter, obtained from Valimet; (b) Cu powders,
<45 mm (325), obtained from Miller; and (c) Fe powders, 45 mm
(325), obtained from Praxair.

66

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

(a)

(b)

(c)

Fig. 6. Microstructure of kinetic sprayed coatings: (a) Al coating on


Al substrate, 200, nozzle inlet pressure 1.7 MPa, inlet temperature
340 C, powder feed rate 65 g/min, stand-off 1.9 cm; (b) Cu coating
on Al, 500, nozzle inlet pressure 1.7 MPa, inlet temperature 590 C,
powder feed rate 113 g/min, stand-off 1.9 cm; and (c) Fe coating on
Al, 200, nozzle inlet pressure 2.0 MPa, inlet temperature 590 C,
powder feed rate 55 g/min, stand-off 1.9 cm.

than that of the powders. There has been a substantial


amount of particle deformation upon impact with the
substrate, leading to relatively low porosity coatings.
For the DOE described in Section 6, we found estimated
porosity ranges as follows: Al, 0.5 to 12%; Fe, 0.1 to
1%; and Cu, 0 to 0.1%.
Results of Vickers indentation hardness measurements
of representative coatings with 100 g loads are given in
Table 1. Measurements and sample preparation on the
pure (bulk) materials were carried out in the same
fashion as on the coatings. This is important because
hardness is sensitive to work hardening and other effects.
Given this sensitivity, one must conclude that the values
for the bulk materials and the corresponding coatings
are perhaps surprisingly close. On the one hand, one
might expect the coatings to have lower hardnesses than
the bulk materials because the coatings are somewhat
more porous. On the other hand, for the kinetic spray
process, with the hard particles striking the substrate,
one might think that the coatings would exhibit more
work hardening than the bulk materials, contributing
to a higher hardness for the coatings. Perhaps these two
effects tend to compensate each other. Note that the
hardness for the Al coating sprayed with He as a carrier
gas is slightly lower than that sprayed with air. This is
surprising, since one would expect higher velocities for
He than for air, and hence denser coatings for the
former. The trend is as expected for Fe, with the Fe
coating sprayed with inlet air at 650 C having a higher
hardness than the one sprayed with inlet air at 430 C.
As argued in Section 3, one would expect the particle
velocities to be higher for higher inlet air temperatures.
Adhesion tests were performed on the coatings using
a tensile pull arrangement. Studs of diameter 0.269 cm
(0.106 in) were glued to the coating and the force
required to pull off the stud was measured. For many
of the coatings, the measured stresses (force divided by
area of the stud) was above 10,000 psi. This is in the
range where epoxies typically fail [10,000 to 12,000 psi
(6882 MPa)]. Failure usually occurred in the epoxy
adhesive, indicating excellent adhesion between coating
and substrate and cohesion within the coatings. Some
Al coatings made during the DOE (Section 6) exhibited
relatively high porosity and relatively poor deformation
of the powder particles upon impact, and these coatings
exhibited substantially poorer adhesion.
Table 2 shows a comparison of oxygen concentrations
as a weight per cent in the powders and in coatings
made from the same powders. These coatings were made
by the kinetic spray process using air as the carrier gas.
Nevertheless, the kinetic spray process has little or no
effect on the oxygen content of the coatings. This is in
marked contrast to thermal spray processes, where, e.g.,
for iron coatings weight percentages of oxygen are
typically in the 210% range. This is presumably because
the kinetic spray powders are at substantially lower

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

67

Table 2
Oxygen concentration in kinetic spray coatings and powders (wt%)

Aluminum
Copper
Iron

Powder

Coatings

0.4%
0.1%
0.1%

0.4%
0.1%
0.2%

temperatures prior to impact with the substrate than


are thermal spray droplets.

5. Coating thickness dependence on inlet air temperature


Figs. 79 show the coating thickness as a function of
inlet air temperature for coatings of Al, Cu, and Fe,
respectively, on substrates of Al, brass, Cu and steel,
with the exception of Cu on steel. Note first that there
is a suggestion of a threshold temperature, below which
the coating material will not adhere to the substrate.
Note this is consistent with a threshold particle velocity
for particle adhesion as observed by Alkhimov and
coworkers [7,8]. As discussed in Section 3, one can
expect that particle velocities will increase roughly as
the square root of the inlet gas temperature. It would
appear from Figs. 79 that the threshold temperatures
are in the range 20050 C. Using the approximate
square root dependence, one finds from Fig. 3 that a
10 mm Cu particle would have a threshold velocity of
about 470 m/s. This compares well with the value of
around 500 m/s measured by Alkhimov and coworkers
[7,8] for a 10 mm Cu particle. Above this threshold
temperature, the coating thickness increases with inlet
air temperature as can be seen in Figs. 79. The dependence on temperature above the threshold appears to
be roughly linear, but there is enough scatter in the data

Fig. 8. Coating thickness versus nozzle inlet gas temperature for Cu


powder ( Fig. 5) sprayed onto Al, brass, and Cu substrates.

Fig. 9. Coating thickness versus nozzle inlet gas temperature for Fe


powder ( Fig. 5) sprayed onto Al, brass, Cu, and steel substrates.

Fig. 7. Coating thickness versus nozzle inlet gas temperature for Al


powder (Fig. 5) sprayed onto Al, brass, Cu, and steel substrates.

that one cannot be sure of the form of the temperature


dependence.
Perhaps we can draw some inferences about kinetic
spray coating mechanisms from this data. In order for
the particle to adhere, the collision with the substrate
cannot be purely elastic. The particle must lose enough
of its kinetic energy that it cannot escape from the
substrate. We saw in Figs. 5 and 6 that there can be
substantial plastic deformation of the particles upon
impact, and this is therefore an available channel for
transference of the particles kinetic energy. The velocity
of the particle would have to be sufficiently high that
its yield stress is exceeded upon collision with the
substrate. This is then a candidate threshold mechanism.
If the kinetic to thermal transition is sufficient to cause

68

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

partial melting of the particles, and this plays a role in


their adhesion, then exceeding the melting temperature
in a localized region is a second candidate for a threshold
mechanism. Another example mechanism would be fracturing of the oxide so that metal/metal adhesive bonding
could occur. It is known [14,15] that metal/metal adhesive bonding can be substantially stronger than
metal/oxide adhesive bonding. One might expect that
the stronger metal/metal bonds might be more likely to
initiate the inelastic mechanisms such as plastic deformation and partial melting. There would presumably be a
threshold velocity for the fracturing of the surface
oxides. There are other mechanisms that one could think
of, and presumably more than one mechanism is at
play. At this time, there is insufficient information to
draw specific conclusions.
From the computational results of the preceding
section, we expect the particle velocities to increase as
the inlet air temperature increases. This, together with
the data of Figs. 79, indicates that the deposition
efficiency increases with the kinetic energy of the particles. The slope of these curves is highest for Al coatings,
next highest for Cu coatings, and lowest for Fe coatings.
This is consistent with either partial melting or deformation mechanisms, since it has ordering inverse to that
of the melting temperatures, yield strengths, hardnesses,
and shear moduli. There is enough scatter in the data
that one cannot tell whether the threshold temperatures
of the three coating materials are different, but presumably they are not identical.
Substrate effects were relatively weak in these experiments. One might expect that the softer substrates would
exhibit lower threshold gas temperatures than the harder
substrates, because the former would be more deformable and hence more likely to promote inelastic collisions
between powder particles and the substrate. However,
the apparent ordering of the threshold inlet air temperatures as a function of substrate hardnesses is different
for different coating materials. Results are sufficiently
close for the different substrates that it is difficult to
draw any conclusions about substrate effects. The one
substrate effect that is confusing is that we were not
able to get Cu to adhere to steel. This is particularly
surprising in view of the observation that Cu coatings
could be readily deposited on the other substrates with
relatively good adhesion and that Fe and Al would
adhere to steel.

substrates. Two values for each of the following four


parameters were chosen to make a 16 run matrix for
each powder on each substrate: air pressure at the nozzle
inlet, air temperature at the nozzle inlet, powder feed
rate, and spray distance or stand-off between the nozzle
exit point and the substrate. The values chosen are listed
in Table 3. Note the maximum inlet air temperatures
chosen are just over half of the melting temperature for
Al and Cu, and well under half for Fe. As seen in Fig. 4,
even the larger particles have temperatures less than the
inlet air temperatures, so the particles are indeed solid
upon impinging the substrate.
Properties measured in the DOE are porosity and
maximum coating thickness. Results are obtained for a
single pass of the nozzle over the target. Because particle
velocities and densities in the spray pattern vary relative
to the center line of the spray pattern, there is a variation
of both thickness and porosity across the single-pass
coating. We found that generally the coating was thickest
and the porosity was least in the center of the pattern
and, correspondingly, in the center of the coating.
Hence, we measured the thickness at the center of the
coatings. Results of the 16 run matrices are available to
those readers who would like to contact the authors. In
the following, some of the more salient results are
summarized.
As noted in Section 4, porosity levels of the Fe and
Cu coatings were relatively low, typically less than 1%.
Porosity values of the Al coatings were more sensitive
to process parameters, however. This may be because
the Al particles are lighter than the Fe or Cu particles.
This is suggested in the porosity profile across the singlepass Al pattern. The porosity values tend to be larger
in the regions far from the center of the pattern than
they are in the center, presumably because the relatively
light Al particles can be swept to the outer regions of
the pattern where the velocities are smaller. It was found
that lower average porosity values for Al were obtained
for the lower inlet air pressure, the lower inlet air
temperature, and the lower powder feed rate. The
powder feed rate effect may be due to the flow and
particle velocities being slowed by overburdening the
nozzle with powder. More data would be needed to
obtain a correlation between process parameters and
coating physical characteristics for Al. The thickness
results to be discussed next indicate that for Al it is
perhaps best to accept some porosity by raising the inlet

6. Design of experiment results

Table 3
Design of experiment parameters

As discussed in Section 2, the apparatus allowed control of process parameters, suggesting a designed experiment to determine how the process depends on the
parameters and their interactions. For these experiments,
Al, Cu, and Fe powders were sprayed onto Al and brass

Powder

Pressure
(MPa)

Temperature
(C )

Powder feed
rate (g/min)

Stand-off
(cm)

Aluminum
Iron
Copper

1.5, 2.0
1.5, 2.0
1.5, 2.0

290, 340
480, 590
480, 590

65, 87
36, 55
73, 113

1.9, 3.8
1.9, 3.8
1.9, 3.8

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

air temperature, because this increases the thickness


substantially.
Next let us look at the DOE results for thickness.
These results are summarized in Fig. 10 in the form of

69

averages over the 16 runs for each powder and each


substrate. This does not provide the two- and higherfactor interactions, but they are available to interested
readers, as noted above. Moreover, these single-factor

(a)

(b)

(c)

(d)

(e)
Fig. 10. Results of the design of experiment (DOE ) for Cu, Al, and Fe coatings on Al and brass substrates: (a) average single-pass thickness; (b)
nozzle air inlet pressure increase; (c) halving gunsubstrate stand-off distance; (d ) nozzle air inlet temperature increase; and (e) powder feed rate.
See Table 3 for values of these parameters.

70

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

averages are substantially more important than the


higher-factor interactions for the DOE results. The
average single-pass thicknesses are similar for the Al
and Fe coatings, but the Cu coatings are substantially
thicker than the other two. Recall that the Cu porosity
values are also the lowest. The mechanical and thermal
properties of Cu elastic modulus, shear modulus,
yield stress, ultimate strength, hardness, melting temperature, etc. are intermediate between those for Al and
Fe. Cu does have the highest Poisson ratio, and although
one can see how the latter could play a role in deformation upon impact, at this time we have to conclude that
it is not clear why the Cu thicknesses are so much
higher. For the Al coatings the most important parameter is the gun (nozzle)substrate stand-off distance. This
appears consistent with Al being relatively light and
hence relatively sensitive to components of velocity
perpendicular to the axis of the nozzle. One might expect
such components to be associated with deflection of the
air flow along the surface of the substrate (target). This
is also consistent with lowering the stand-off distance
having a positive but smaller effect for Cu and Fe. The
second largest effect on the Al coating thickness was
due to a nozzle air inlet temperature increase. This is
understandable, since such temperature increases lead
to higher particle velocities which, in turn, foster greater
plastic deformation of the splats. As discussed above,
such inelastic effects are necessary for particles to stick
and can therefore lead to higher deposition efficiency
and thickness. The temperature effect is also strong for
Fe coatings, but is relatively weak for Cu coatings. In
fact, the Cu coatings were sensitive only to powder feed
rate, with the possible exception of the stand-off distance
for Cu on brass. Perhaps this insensitivity is because Cu
deposits so easily for, again, reasons to be determined.
Fe coating thicknesses are aided somewhat by pressure
increases, although for Cu and Al pressure changes are
the weakest effect. Substrate material has a relatively
small effect in all cases, as can be seen in Fig. 10.

7. Conclusions
Our new high-velocity spray apparatus has provided
the capability of controlling process parameters important to the kinetic spray process. This process forms
coatings via conversion of particulate kinetic energy to
mechanical and thermal deformation of particles upon
impact with a substrate. Coatings were produced by
accelerating solid particles of Al, Cu, and Fe onto metal
substrates. These coatings were found to have relatively
low porosity values, hardnesses comparable with the
corresponding bulk materials, adhesive strengths as high
as 6882 MPa, and oxide contents essentially the same
as in the powders.
Our computations suggest that our nozzle obtains air

velocities over 1000 m/s, and the 1 mm Cu particles


follow the air velocities rather closely, while the larger
Cu particles are slower. The larger Cu particle temperatures approach that of the inlet air upon nozzle exit,
while the smaller particles are cooler.
The new high-velocity spray apparatus has provided
the capability of controlling process parameters important to the kinetic spray process. In studies of coating
thickness as a function of nozzle inlet air temperature,
a threshold behavior was observed. That is, the powders
essentially do not stick to the substrates below a certain
inlet air temperature range. This is consistent with
several inelastic mechanisms such as plastic deformation
and partial particle melting. Above this threshold, the
deposition efficiency increases with inlet air temperature,
consistent with increasing particle velocities. Substrate
effects appeared to be relatively weak in this experiment
and in our subsequent designed experiments (DOE).
In the DOE the nozzle air inlet pressure, inlet air
temperature, nozzlesubstrate stand-off distance, and
powder feed rate were varied. Al, Cu, and Fe powders
were sprayed onto Al and brass substrates. The quantities we measured were porosity and maximum coating
thickness in a single pass. Porosity values were found
to be less than 1% for Cu and Fe, but in the case of Al
porosity was sensitive to the spray parameters. Al and
Fe had similar average single-pass thicknesses, while Cu
thicknesses were much higher. In fact, Cu thicknesses
were sensitive primarily only to powder feed rates. Al
coatings were sensitive to spray parameters, presumably
because Al is the lightest of the three materials and
hence sensitive to air velocity components perpendicular
to the axis of the nozzle. Consistent with this, the most
important parameter is the nozzlesubstrate stand-off
distance, perhaps due to deflection of air flow along the
substrate surface. Increasing nozzle inlet air temperature
also had a positive effect on Al and Fe coatings, as
expected because this increases particle velocities.
Inelastic processes are essential for particle deposition,
and the DOE results are consistent with plastic deformation and partial melting processes.
More work is needed to determine the relative importance of these processes and indeed to investigate other
possible processes, however. Moreover, we need further
experiments to understand the differences we observed
in the spray behavior of Cu relative to that of Fe and Al.

Acknowledgement
The consortium which carried out this research was
organized under the auspices of the National Center for
Manufacturing Sciences. Flame Spray Industries
designed and assembled the equipment and the research
facility is located at the GM R&D Center. Noel Potter
measured the oxygen in the coatings. Pat MacDonald

T.H. Van Steenkiste et al. / Surface and Coatings Technology 111 (1999) 6271

provided valuable technical assistance. The authors


are grateful for partial support provided by the
Manufacturing Technology Directorate, Wright
Laboratory ( WL/MTX ), Air Force Material Command,
USAF under Cooperative Agreement Number
F33615-94-2-4423. The US Government is authorized
to reproduce and distribute reprints for Governmental
purposes notwithstanding any copyright notation
hereon. The views and conclusions contained herein are
those of the authors and should not be interpreted as
necessarily representing the official policies or endorsements, either expressed or implied, of Wright Laboratory
or the US Government.

References
[1] R.A. Miller, J. Thermal Spray Technol. 6 (1997) 35.
[2] R.C. McCune, Welding J. 24 (1995) 41.
[3] G.H. Smith, N.Y. Kenmore, R.C. Eschenbach, J.F. Pelton, US
Patent 2,861,900, 25 November 1958.

71

[4] C.F. Rocheville, US Patent 3,100,724, 13 August 1963.


[5] J.A. Browning, in: C.C. Berndt ( Ed.), Proc. United Thermal
Spray Conf., ASM International, Indianapolis, September 1997.
[6 ] J.A. Browning, US Patent 5,271,965, 21 December 1993.
[7] A.P. Alkimov, V.F. Kosarev, A.N. Papyrin, Dokl. Akad. Nauk
SSSR 318 (1990) 1062.
[8] A.P. Alkhimov, A.N. Papyrin, V.F. Kosarev, N.I. Nesterovich,
M.M. Shushpanov, US Patent 5,302,414, 12 April 1994.
[9] R.C. McCune, A.N. Papyrin, J.N. Hall, W.L. Riggs II, P.H.
Zajchowski, in: C.C. Berndt, S. Sampath ( Eds.), Proc. 8th
National Thermal Spray Conf., ASM International, Houston,
September 1995.
[10] R.C. McCune, W.T. Donlon, E.L. Cartwright, A.N. Papyrin, E.F.
Rybicki, J.R. Shadley, in: C.C. Berndt (Ed.), Thermal Spray:
Practical Solutions for Engineering Problems, ASM International,
Materials Park, OH, 1996.
[11] J.D. Anderson Jr., Modern Compressible Flow, McGraw-Hill,
New York, 1982.
[12] C.B. Henderson, AIAA J. 14 (1976) 707.
[13] D.J. Carlson, R.F. Hoglund, AIAA J. 2 (1964) 1980.
[14] G.L. Zhao, J.R. Smith, J. Raynolds, D.J. Srolovitz, Interface Sci.
3 (1996) 289.
[15] T. Hong, J.R. Smith, D.J. Srolovitz, Acta Metall. Mater. 43
(1995) 2721.

Das könnte Ihnen auch gefallen