JTTEE5 8:565-575

ASM International

Peer Reviewed
Effect of Nozzle Configuration,
Gas Pressure, and Gas Type on Coating
Properties in Wire Arc Spray
X. Wang, J. Heberlein, E. Pfender, and W. Gerberich

(Submitted 15 Oct 1998; in revised form 19 May 1999)

Wire arc spraying is a coating process in which minor modifications of the configuration and spray pa-
rameters can have a strong impact on coating characteristics. A study on the effects of the fluid dynamics
of the atomizing gas on the coating properties is presented. Different types of nozzles, shrouds, and gases
have been used to provide various flow velocities and reactive environments in the metal atomization re-
gion. The effects on particle velocity, coating density, composition, and interface characteristics between
the coating and the substrate have been evaluated. It is clear that higher gas velocities improve practi-
cally all coating properties, but also increase oxide content in the coating. However, the oxidation can be
drastically reduced if nonoxidizing gases are used for atomization in combination with a shroud. A dis-
cussion on the physical effects contributing to the observed adhesion improvements and interfacial prop-
erties is provided.

bore, B is a converging-diverging nozzle allowing supersonic

Keywords adhesion, coatings, nozzles, porosity, wire arc spraying
flow of the atomizing gas to extend farther into the atomizing re-
gion, C is a standard nozzle with secondary gas injection, and D
1. Introduction
Wire arc spraying is an efficient and economical coating
technique, which has a wide range of applications in industry
(Ref 1-3). Some of its principal attractions are its operational
simplicity and the low cost for the wire feedstock. However, this
simplicity does not exclude that minor modifications of the
equipment design features can strongly influence the coating
properties. Because of the resurgent interest in this process, a
number of new developments in improving the equipment have
been initiated and introduced. This is particularly true for the
way the atomizing gas is delivered to the wire tips. In the wire
arc spray process, the arc between the tips of two continuously
advancing wires melts the material, and the molten metal is re-
moved and blown toward the substrate by a high-velocity gas
stream. This gas stream is responsible for forming the initial
droplets after removal of the molten material from the wire tips
(primary breakup), for atomizing the larger droplets (secondary
breakup), and for accelerating them toward the substrate. Higher
gas velocities will result in the generation of finer droplets,
which are accelerated to higher velocities. A detailed descrip-
tion of these effects has been presented by Steffens (Ref 4).
In this article, the results of a systematic investigation of the
influence of different nozzle and shroud configurations on the
coating properties is presented, including the effects on particle
size distribution, particle velocities, coating microstructure and
composition, and coating adhesion. The nozzle configurations
investigated are shown schematically in Fig. 1 and can be char-
acterized as follows: A is the standard nozzle with a straight

X. Wang, J. Heberlein, E. Pfender, and W. Gerberich, ERC for

Plasma-Aided Manufacturing, University of Minnesota. Contact e-
mail: [email protected]. Fig. 1 Schematic of nozzle configurations

Journal of Thermal Spray Technology Volume 8(4) December 1999565

 represents the addition of a shroud to the nozzle with secondary Metco U.S. Inc., Westbury, NY) 4RG gun was used; Table 1 lists
Peer Reviewed

gas injection. Each of these arrangements has different fluid dy- the operating conditions. For all the cases, the arc voltage, cur-
namic features, and results obtained with each of them are pre- rent, and standoff distance were set at 30 V, 150 A, and 15 cm,
sented. respectively. In cases A and B, aluminum was sprayed onto a
steel substrate, while in cases C and D, stainless steel was
sprayed onto an aluminum substrate.
2. Experimental Methods
2.2 Measurement of Gas Velocity and Particle
2.1 Process Conditions Velocity
Before spraying, the substrates were treated by grit blasting, A Pitot tube was used to measure the gas velocity at the noz-
acetone degreasing, and ultrasonic cleaning. A Metco (Sulzer zle exit. In incompressible flow, the Bernoulli equation can be
used to relate changes in velocity and pressure along the stream-
Table 1 Operating conditions line when the flow is not affected by friction:
Nozzle
1
configuration Gas pressure, psi Gas flow rate, scfm Gas type P0 = P +   ρ V 2
A 45, 65, 85 9, 13, 17 Air 2
B 65 15 Air
C Primary: 60 Primary: 12 Air where P0 is the stagnation pressure, P is the static pressure at a
Secondary: 15 Secondary: 3 Air point in the flow where the velocity is V, and ρ is the gas density.
D Primary: 60 Primary: 12 Carbon Thus, measurement of the stagnation pressure and the static
Secondary: 15 Secondary: 3 dioxide, pressure will yield the gas velocity (Ref 5).
nitrogen
Images of particle streaks were taken with a Kodak high-
speed vision system (Eastman Kodak, Rochester, NY). The ex-
posure time was adjusted until clear particle streaks could be
recorded. The streak length was then determined using image
analysis and divided by the exposure time, thus giving particle
velocity. About 40 images were evaluated to obtain average par-
ticle velocities. For these measurements, using a diaphragm re-
duced the metal droplet density in the stream. Figure 2 shows the
experimental setup for the particle velocity measurements.

2.3 Gas Stream Visualization

A shadowgraph setup, as shown in Fig. 3, was used to study
the gas flow (Ref 6). A mercury arc lamp with a 2 mm diameter
aperture was the point light source. The light was collimated us-
Fig. 2 Experimental setup for particle velocity measurements ing a 21.6 cm diameter spherical mirror with a focal length of
168 cm. A plane mirror turned the light by 90° to pass through
the flow. The light was again turned 90° by a plane mirror and fo-
cused to a point by an identical spherical mirror. A Vivitar (Vivi-
tar Corp., Newbury Park, CA) zoom lens with a fully open
aperture was used to focus the light onto the film. A shutter
speed of 1⁄1000 s and 400 ASA black and white film were used to
produce the images shown in Fig. 7 (Ref 6).

2.4 Coating Microstructure and Composition

Analysis
Micrographs of polished coating cross sections were investi-
gated using reflected light microscopy with a green filter for im-
proved contrast at 200× magnification. Scanning electron
microscopy (SEM) was used to examine the cross sections of the
samples and microfractography. The surface morphology of
coatings was also studied by SEM. Energy dispersive x-ray
analysis was used for determining composition of the interfacial
region and fracture surfaces. Auger electron spectroscopy is a
nondestructive technique with a sensitive depth resolution suit-
able for elemental analysis of materials. To eliminate surface
Fig. 3 Experimental setup for shadowgraph measurement contamination, the sample surface was etched to a depth of 300

566Volume 8(4) December 1999 Journal of Thermal Spray Technology

Å by argon ion beam sputtering. The atomic concentrations Droplet Formation from the Wire Tips and Droplet Size

Peer Reviewed
were calculated from the intensity of the Auger peaks based on Distributions. Figure 4 shows a high-speed photograph of an
a calibration of relative sensitivity. Chemical composition pro- asymmetric melting behavior of the cathode and the anode wire.
files were taken using a scanning Auger microprobe to study the The anode melts slowly, resulting in elongated relatively large
film-substrate interfacial chemistry. On a perpendicular section droplets. This uneven melting leads to an asymmetry of the arc
of the coated sample, a linear scan along the direction orthogo- and affects the spray pattern and the coating structure. At the
nal to the boundary surface was taken to determine concentra- cathode, melting is more localized, and the molten droplets are
tion profiles of the elements in the film and substrate. X-ray immediately blown away by the atomizing gas flow, resulting in
diffraction with CuKα radiation was used for identification of relatively small droplets. The higher melting rate at the cathode
the intermediate phases formed at the interfacial region and ox- is due to a more constricted arc attachment compared to the more
ide phase of coatings. diffuse arc attachment at the anode. From sequences of high-
speed videographs, it is apparent that higher atomizing air pres-
2.5 Porosity and Particle Size Measurement sures result in smaller droplet sizes. This fact has been
corroborated by measurement of the particle size distributions,
Image analysis was used to determine the porosity of pol- shown in Fig. 5.
ished coating cross sections. The analysis used differences in Oxide Content of the Coating. Arc sprayed metal coatings
gray levels to distinguish different features of the coating micro- contain a certain amount of oxides. A quantitative analysis using
structure. Features such as porosity were detected through their Auger electron spectroscopy (AES) revealed that the amount of
respective gray scale ranges, and pore sizes were measured. The aluminum oxide in the coating increased with increasing atom-
porosity measurements were performed for 12 different loca- izing air pressure. The amount of oxide content calculated from
tions randomly distributed over the coating cross sections to ob- the peak intensity of the AES in the coatings has been 18.2
tain average results. Droplet size was determined by spraying (±1.6), 19.5 (±1.9), and 24.8% (±2.1%) as the gas pressure in-
into ice and observing the resulting solidified particles with
creased from 0.31 to 0.45 MPa and to 0.59 MPa, respectively.
SEM combined with image analysis.
During spraying, the effect of atomizing air and the entrainment
of the surrounding air into the spray stream caused significant
2.6 Adhesive Bond Strength Test inflight oxidation of the molten metal particles (Ref 11). Increas-
ing the atomizing air pressure leads to higher gas stream veloci-
The bond strength of coatings was measured in accordance
ties, which in turn break up the molten particles into smaller
with the ASTM C 633-79 standard pull-off tensile test. This is a
droplets. The smaller droplets react more readily with oxygen
common method of characterizing the comparable bond
than the larger droplets, because of their greater specific surface
strength of thermally sprayed coatings. The results of the tests
area. Thus, coatings sprayed under higher atomizing air pressure
determine the degree of adhesion of a coating to a substrate in
tend to exhibit a higher oxide content.
tension normal to the surface. Twelve tests were performed on
each sample using an Instron machine (Instron Corp., Canton, Coating Porosity. Figure 6 shows polarized optical mi-
MA) at a crosshead speed of 0.075 cm/min. crographs of cross sections of sprayed coatings. Using comput-
erized image analysis, the porosity of the coatings has been
determined quantitatively. The porosity of the coatings is 27
2.7 Microhardness Test (±4), 18 (±4), and 12% (±3%) for gas pressures of 0.31, 0.45,
A Vickers microhardness indenter was employed to measure and 0.59 MPa, respectively, indicating that the gas pressure is an
Vickers hardness profiles across the coating-substrate interface. important parameter in terms of porosity. The sprayed coating is
Measurements were made using a load of 200 g for 10 s. On each built up particle by particle and, therefore, higher atomizing air
sample, twelve measurements were made at random locations. pressure results in higher impact velocity of smaller particles on
the substrate. Faster molten particles with higher kinetic energy
spread and deform more readily on impact, thus increasing coat-
3. Results and Discussion ing density and reducing porosity.

3.1 Effect of Atomizing Gas Pressure and Flow

Rate on Coating Properties with the Straight
Bore Nozzle
Arc fluctuations due to periodic removal of molten droplets
from the wire tips by the atomizing gas stream are characteristic
for the wire arc spraying process (Ref 2, 4, 7-10). Dynamic
forces of the atomizing air stream acting on the arc as well as on
the droplets lead to different particle sizes, which, in turn, deter-
mine porosity and oxide content of sprayed coatings. In this
study, arc spraying of aluminum coatings was considered with
the objective of establishing correlations between atomizing air
pressure and coating properties such as porosity and oxide con-
tent. Fig. 4 High-speed image of droplet formations

Journal of Thermal Spray Technology Volume 8(4) December 1999567

Peer Reviewed

(a)

(a)

(b)

(b)

(c)

Fig. 5 Aluminum particle size distributions. (a) 0.31 MPa. (b) 0.45
MPa. (c) 0.59 MPa

3.2 Effect of Converging-Diverging Nozzle on

Coating Properties (c)

Fig. 6 Micrographs of coating cross sections. (a) 0.31 MPa. (b) 0.45
To produce coatings with high density and high bond MPa. (c) 0.59 MPa
strength, a converging-diverging atomizing gas nozzle (CD
nozzle) was used to accelerate the gas jet to supersonic levels atomizing gas speeds to supersonic velocities by properly ex-
without choking, thus increasing particle velocity (Ref 12). To panding the flow in a converging-diverging nozzle as opposed
achieve better corrosion and wear resistance, coatings with to the conventional straight bore nozzle is expected to improve
minimal porosity and maximum strength are required. However, coating density and adhesion. The converging-diverging nozzle
the velocities of the particles are subject to certain limitations. is shown schematically in Fig. 1(b). The nozzle throat-to-exit
Coatings produced with conventional nozzles have relatively area ratio is matched to the required pressure ratio for the condi-
high porosity and relatively low bond strength. Increasing the tions of Mach number 1.5.

568Volume 8(4) December 1999 Journal of Thermal Spray Technology

 Peer Reviewed
(a) (a)

(b)
(b)
Fig. 7 Shadowgraphs of gas flow pattern. (a) Conventional straight Fig. 8 Particle size distributions. (a) Conventional straight bore noz-
bore nozzle. (b) Converging-diverging nozzle zle. (b) Converging-diverging nozzle

Flow Patterns of Atomizing Gas Stream. Figure 7 shows CD nozzle and a conventional nozzle. It is obvious that the flat-
shadowgraphs of gas streams produced by the CD nozzle and a tening behavior is different for the two cases. The molten parti-
conventional nozzle. Visualization of the atomizing gas stream cles sprayed with the CD nozzle spread out to a much greater
produced with a straight bore nozzle illustrates the formation of extent than those sprayed with the conventional nozzle. They
a strong shock structure and nonuniformity. This effect causes also show a higher degree of deformation, and there are fewer
rapid velocity decay. In contrast, the gas stream produced with a unmolten droplets at the edges of the flattened particles. Figure
CD nozzle has a very weak shock structure and a visually longer 10 shows cross-sectional views of aluminum coatings sprayed
high velocity potential core. Because the gas stream drives the with the CD nozzle and the conventional nozzle illustrating that
molten particles, the associated molten particle velocities are in- the coatings prepared with the CD nozzle have lower porosity
creased as well. than those produced with the conventional nozzle. Also, the
Particle Size Distribution. Figure 8 shows particle size dis- pore sizes of the coatings obtained with the CD nozzle are in
tributions for two different nozzles. For both conditions the par- general smaller than those of the coatings obtained with the con-
ticles have a bimodal size distribution. This distribution is likely ventional nozzle.
the result of the difference in size of the initial droplets from the The higher velocity of the atomizing gas stream obtained
anode and the cathode. But it is obvious that the higher gas ve- with a converging-diverging nozzle readily explains these ob-
locities of the converging-diverging nozzle lead to smaller par- servations. In wire arc spraying, the particle has its highest tem-
ticles with a narrower distribution. perature in front of the wire electrode, and a higher atomizing
Coating Microstructure. The particle velocity and the par- gas velocity will lead to higher particle acceleration. The conse-
ticle temperature determine the coating structure at the instant of quence is not only a higher impact velocity, but also a higher
impact on the substrate. Completely molten particles impinging particle temperature at impact because of the shorter flight
on the substrate spread out radially in the form of thin disks. In time and less cooling by entrained air. The higher tempera-
reality, however, the deposit is not uniform in thickness, and the tures result in lower viscosities and better wettabilities, yield-
periphery of the flattened particle is not circular. Figure 9 shows ing better conformal coatings in the case of using the CD
the surface morphology of aluminum coatings prepared with the nozzle (Ref 13).

Journal of Thermal Spray Technology Volume 8(4) December 1999569

Peer Reviewed

(a) (a)

(b)
(b)
Fig. 10 Micrographs (200×) of cross sections of a coating under two
Fig. 9 Surface morphology of coatings. (a) Conventional straight conditions. (a) Conventional straight bore nozzle. (b) Converging-di-
bore nozzle. (b) Converging-diverging nozzle verging nozzle

Coating Properties. A significant criterion for the quality of lates with the large phase heterogeneity and porosity of the coat-
thermally sprayed films is their adhesion to the substrate. The ings.
coating is built up particle by particle. Molten particles undergo
severe deformation and rapid solidification when they impinge
on the substrate. To achieve high bond strength between film 3.3 Effect of Secondary Gas or Inert Gas on
and substrate, the particles must be in a fully molten state and Coating Microstructure
have sufficient velocities to be able to spread out and flow into Coating density and bond strength depend to a large degree
the contours and crevices of the roughened substrate. The results on particle velocities. In conventional wire arc spraying, the ve-
of adhesion tests indicate that the bonding strength of the alumi- locities of the larger particles are relatively low, limiting the
num coating sprayed with the CD nozzle (about 35 MPa) is bond strength of the coating. Spraying with secondary gas at-
higher than that sprayed with the conventional nozzle (about 20 omization results in more uniform particle size distributions,
MPa), as shown in Table 2. more focused spray patterns, higher particle velocities, and im-
Microhardness is of interest because it gives an indication of proved coating properties. A modified nozzle with secondary
resistance to abrasive wear. It is greatly affected by the amount gas injection as shown in Fig. 1(c) was used to optimize atomi-
of oxide and porosity of the deposit. The coating with more zation and enhance particle velocities. A primary axial air
pores has less resistance to penetration than the denser coating, stream removes the molten metal droplets from the wire tips and
consequently displaying lower hardness. Conversely, oxide par- from the area of the wire intersection, and a secondary air stream
ticles in the coating tend to give a higher hardness. The values forms a conical sheath around the axial air stream. The primary
for the microhardness cover a relatively wide range. This corre- gas stream and the secondary gas stream emerge from the nozzle

570Volume 8(4) December 1999 Journal of Thermal Spray Technology

 Peer Reviewed
(a) (b)

Fig. 11 Particle size distributions of stainless steel particles atomized with CO2. (a) Nonshrouded. (b) Shrouded

Table 2 Bond strength and microhardness of aluminum coatings from different nozzles (12 measurements)
Nozzle Bond strength Bond strength Microhardness Microhardness
type (mean), MPa (σ), MPa (mean), HV (σ), HV
Converging-diverging nozzle 35.4 7.1 59 11
Straight bore nozzle 20.2 6.2 51 9

as coaxial gas streams, thus tending to protect the droplets from air atomization only (Fig. 1a). The atomization of the molten
entrained air and to concentrate the flow pattern of the droplets. droplets is enhanced by secondary gas injection, which leads to
Lower coating porosity and higher bond strength are expected smaller droplets, which are accelerated faster by the gas flow.
from the secondary gas injection. Results of image analysis of high-speed photographs of particle
Air atomization is commonly used in the wire arc spray proc- flight trajectories indicate that in the case of secondary gas at-
ess. The major advantages are the availability and economy of omization, the average particle velocity is about 105 m/s (±10
compressed air. In the air atomization wire-arc spray process, m/s) when particles impact the substrate surface, while in the
the oxide content of the sprayed coating is relatively high due to case of primary gas atomization, the particle velocity is about 70
oxidation of the molten wire material. This higher oxide content m/s (±8 m/s).
can increase the coating hardness so that the abrasion and wear Particle Size Distributions. Droplet size was determined by
resistance of the coatings is improved. However, the oxide con- spraying into ice and analyzing the size distribution of the solidi-
tent may also be detrimental to coating properties because ox- fied particles with SEM and image analysis system. Assuming
ides may reduce the adhesion strength between coating and that all particles are captured by ice, the size distributions for
substrate. Also, hard oxide particles embedded in sprayed coat- particles sprayed with CO2 as atomizing gas with and without
ings impose problems during machining. Furthermore, coatings shroud are shown in Fig. 11. There is a clear reduction in average
sprayed with air atomization often contain relatively high poros- particle size, but a bimodal distribution is observed in both
ity, which is frequently detrimental. Another disadvantage of air cases. The smaller particles probably originate from the cath-
atomization is related to the burnoff of alloying elements (such ode, which has a more constricted attachment and higher atomi-
as chromium and carbon) contained in parent wires. These ele- zation.
ments are essential ingredients to produce the required coating Coating Microstructure. Figure 12 shows scanning elec-
characteristics. As a consequence, coatings with specified char- tron micrographs of cross sections of sprayed coatings. Using
acteristics cannot be produced reliably. In this study, nitrogen computerized image analysis, the porosity of the coatings was
and carbon dioxide have been used as atomizing gas in wire arc determined quantitatively. The porosity in the case of air atomi-
spraying to produce coatings with higher quality. zation was 17% (±3%), which is higher than that in the case of
However, even spraying with nitrogen and carbon dioxide as CO2 atomization (12 ± 2%) and N2 atomization (13 ± 2%). The
atomizing gas may still result in strong oxidation of coatings be- porosity in the case of gas shrouding was substantially reduced
cause of entrainment of large amounts of air (Ref 11). Air en- to 6% (±2%) for CO2 atomization and 8% (±2%) for N2. This
trainment can cause a strong drop of temperature and velocity of can be explained by two factors: (a) The shrouded inert gas
the gas stream. An obvious approach to avoid air entrainment is sprayed coatings result in the lowest oxide content; therefore,
to shield the atomizing gas flow from the surrounding atmos- there is a more uniform contact between metallic splats, which
phere by using a shroud. makes the coating more consistent and less porous. (b) Because
Gas Velocity, Particle Size, and Particle Velocity. The the shrouded nozzle with secondary inert gas injection leads to
measurements show that the gas velocity (610 ± 22 m/s) at the higher acceleration of the droplets and less cooling by the en-
nozzle exit in the case of primary/secondary air atomization trained air, the coating was built up by particles with higher ve-
(Fig. 1c) is higher than that (530 ± 15 m/s) in the case of primary locity and temperature leading to the lowest porosity. Defects

Journal of Thermal Spray Technology Volume 8(4) December 1999571

 were hardly seen in the deposited layer, and the structure was nated from oxide/metal interfaces due to the large amount of ox-
Peer Reviewed

very fine when using shrouded nozzles with secondary inert gas ide embedded inside the coating. In the case of inert gas atomi-
atomization. In air sprayed coatings, many microcracks origi- zation, particularly in the case of shrouded inert gas atomization,
fewer microcracks of the same size were observed. It is possible
that there were more smaller microcracks, which have not been
resolved, under these conditions. However, for the same SEM
with the same magnification, more cracks were observed when
using air sprayed coatings than shrouded inert gas sprayed coat-
ings. This can be explained by the fact that less oxidation yields
better conformal coatings.
Alloy Element Loss. Figure 13 shows the chromium content
of the original stainless steel wire, and of the air, CO2, and N2
sprayed coatings tested by AES. The results indicate that when
air is used as the atomizing gas, a significant amount of chro-
mium is lost, as shown by the fact that the deposited coating con-
tained 11 wt% (±2%) Cr compared with the wire having 19 wt%
(±2%) Cr. The formation of CrO3 by the reaction: Cr2O3 (s) +
3 ⁄ O (g) → 2 CrO (g) becomes significant at high tempera-
2 2 3
tures (Ref 15), resulting in chromium loss by evaporation of
CrO3. Using a shrouded nozzle with secondary CO2 or N2 gas
injection, the chromium content of the coating was essentially
(a)
the same as that of the original wire. This indicates that the
shrouded nozzle reduces the mixing of entrained air with the at-
omizing gas.
Oxide Content of the Coating. The amount of oxide de-
pends on both the amount of oxygen present in the spraying jet
and the total surface area of the particles in the jet. Figure 14
shows the oxide content. The amount of oxide in the air sprayed
coating (21%) was higher than that of the inert gas sprayed coat-
ing, and the use of the shrouded nozzle with inert gas further de-
creased the oxide content. Using inert gas without a shrouded
nozzle does not eliminate the oxides because of the entrainment
of air from the surroundings (Ref 11). However, using a
shrouded nozzle with secondary inert gas injection reduced the
oxide content even though the particle size was decreased. This
indicates that using a shrouded nozzle can result in coatings with
both higher density and less oxide content.

(b)
3.4 Adhesion
Fig. 12 Micrographs of coating cross sections. (a) Nonshrouded. (b)
Shrouded Section 3.3 presents results showing narrower size distribu-
tions and denser coatings when secondary gas injection was
used. Thus, higher bond strength is expected to be obtained with

Fig. 13 Chromium content of stainless steel wire and coatings

sprayed with different gases Fig. 14 Oxide content of coatings sprayed with different gases

572Volume 8(4) December 1999 Journal of Thermal Spray Technology

secondary gas atomization. In this section, the effect of secon- mass convection, and interfacial reaction between coating and

Peer Reviewed
dary gas injection on adhesion is evaluated. substrate form an interfacial transition zone. Small amounts of
Sprayed coatings are formed by the impact, deformation, and solid solutions and intermetallic compounds formed at the inter-
rapid solidification of individual molten droplets so that their facial region, which improved adhesion between coating and
structure consists of a series of overlapping lamellae. The adhe- substrate.
sion of the coating depends upon the interactions between indi- Bond Strength. The results of tensile adhesion tests (Table
vidual lamellae and between lamellae and substrate (Ref 16). 3) indicate that the bond strength of the coating by secondary gas
The bond strength of a coating is affected by the extent of both atomization is 40 MPa, while that by primary gas atomization is
physical and chemical interactions between the coating and the only 28 MPa.
substrate material and on the microstructure of the interfacial re- To reveal the fracture mechanisms of the sprayed coatings,
gion. Poor adhesion can be attributed to poor interfacial inter- photographs of fracture surfaces were obtained by SEM. In the
locking, low degree of metallurgical bonding, and high internal case of primary gas atomization, the fracture occurred almost
stresses. The degradation modes of the coating depend on both entirely along the weakly bonded interface between the coating
the nature of the coating-substrate interface and on the chemical and substrate. The fracture surface was relatively smooth. It dis-
phenomena that occur at the interface during deposition and played a quasi-cleavage fracture pattern. Coexistence of iron
utilization. Therefore, a detailed interface microanalysis could and aluminum peaks in EDS spectra reveal that the fracture
yield interesting information on the role played by interfacial mode is failure in the interfacial region. In the case of secondary
chemistry in coating-substrate adhesion. gas atomization, fracture occurs within the sprayed film. The
The methods of testing coating adhesion of thermally photograph of the fracture surface shows a dimple fracture pat-
sprayed coatings have been discussed by several other investi- tern, indicating ductile failure. Cracks propagate mainly along
gators (Ref 17-20). Even though pull-off tensile testing has unbonded interparticle contact areas and through regions of
some limitations, it is acceptable to obtain comparable bond stress concentration such as pore edges and oxide inclusions
strengths of thermally sprayed coatings. In this study, the bond within the sprayed film. The only iron peak in the EDS spectra
strength of coatings produced with both nozzles was measured confirms that the failure mode of the sample obtained by the sec-
by pull-off tensile tests according to ASTM C 633-79. The inter- ondary gas atomization is cohesive fracture within the steel
facial regions were structurally and compositionally studied by coating.
SEM, energy dispersive x-ray spectrometry (EDS), and AES to Microstructure and Chemistry of Interface. Figure 15
obtain relationships between structure and adhesion. The results shows element concentration profiles of coating/substrate cross
of the experiments showed that the bonding mechanism of coat- sections. In the case of primary gas atomization, there is a clear
ing sprayed with secondary gas atomization is not only me- boundary between coating and substrate, while in the case of
chanical but also metallurgical. Figure 15 shows a transition secondary gas atomization, there exists an intermixed transition
region between coating and substrate. This region was caused zone between deposited coating and substrate. The formation of
by metallurgical bonding. Localized solid interdiffusion, molten this intermixed zone can be attributed to metallurgical bonding

Table 3 Bond strength and microhardness of steel coatings from different atomizations (12 measurements)
Atomization Bond strength Bond strength Microhardness Microhardness
type (mean), MPa (σ), MPa (mean), HV (σ), HV
Primary/secondary gas atomization 40.2 6.9 396 15
Primary gas atomization 28.4 5.1 352 13

(a) (b)
Fig. 15 Element concentration profiles across coating/substrate. (a) Primary gas atomization. (b) Primary/secondary gas atomization

Journal of Thermal Spray Technology Volume 8(4) December 1999573

 resulting from molten mass convection, solid interdiffusion, and similarity in d-spacing for certain peaks of the iron and Fe3Al
Peer Reviewed

intermediate phase formation. phases. A further difficulty for identification of iron-aluminum

In the case of secondary gas atomization, the sprayed parti- is the peak broadening and peak shifting that occur due to lattice
cles are of small size and mass, have considerably higher veloci- distortion by oversaturated solid solution and thermal stress. Re-
ties, and have higher temperatures according to the situation in action between coating and substrate invariably forms equilib-
the spray jet. They are drastically deformed when striking the rium phases. At the interfacial region of the secondary gas
substrate surface and have a high cooling rate. The strongest ad- sprayed steel-aluminum sample, either solid solution or com-
hesion produced by secondary gas atomization is based on met- pound formation takes place readily, reaching the lowest free en-
allurgical interaction between the deposited film and the ergy composition. The simplest reaction is the formation of a
pretreated substrate. Using secondary gas atomization, particles solution of one phase in the other, leading to and maintaining
still remain in the molten state at the instant of impingement on equilibrium saturation in both phases at the interfacial region. A
the substrate because of the short flight time. Therefore, the sur- continuation of the reaction is associated with the formation of
face of the aluminum substrate will locally melt by heat transfer intermediate phases. The intermetallic compounds formed dur-
from the molten particles, and mixing among the liquid phases ing reactions are compatible with both phases at the coating and
may occur. Thus, molten substrate metal can infiltrate into po- the substrate. Several recent studies have shown that the me-
rous regions of the deposit. Meanwhile, molten particles can chanical properties of heterophase interfaces can be controlled
also flow into the cavities of the molten substrate. Successive by interfacial reactions (Ref 21, 22). Introducing either inter-
impingement of molten particles enhances the fluid dynamic metallic compounds or solid solutions in the transition zone can in-
mixing. This type of molten mass convection results in localized crease the strength of the interface between coating and substrate.
microwelding of the coating and substrate, which contributes to Adhesion Mechanisms. Secondary gas sprayed stainless
metallurgical bonding. steel coatings on aluminum substrates reveal higher bonding
When molten particles impinge on the substrate, they spread strength than those sprayed with only primary gas. Failure oc-
out, cool down, and solidify at an extremely high quenching curs preferentially within the deposited film due to its stronger
interfacial strength. The following three types of bonding mecha-
rate. The diffusion process which, because of the very short time
nisms are operative in spraying with secondary gas atomization:
scale, cannot be described by the classical concepts of diffusion,
can be interpreted to be due to the fact that the surface of the sub- • Physical bonding: The action of van der Waals forces be-
strate has been markedly disturbed by preparatory grit blasting tween coating and substrate
and hence exhibits a high defect concentration. Thus, the molten
particles with high temperatures exhibit a high vacancy concen- • Mechanical bonding: A molten particle striking a rough-
tration. Due to this anomalous condition, the interdiffusion ened substrate surface, given that it has sufficient fluidity,
process by vacancy-atom interchange can occur with higher dif- will assume the surface topography. Mechanical interlock-
ing between the protrusions of the deposit and the rough-
fusion coefficients than is normally the case. Thus, solid inter-
ened substrate leads to mechanical adherence. Higher ve-
diffusion also contributes to metallurgical bonding. locity and temperature of the particles in secondary gas
Also, intermediate phases are produced during the reaction atomization enhance mechanical bonding.
simultaneously with the diffusion. Energy dispersive x-ray
analysis has been carried out on the entire transition zone of the • Metallurgical bonding: In secondary gas atomization, ad-
sample cross section, which was deposited by secondary gas at- hesion improvement could be due to metallurgical bonding
omization. Results indicate that there is a strong elemental mix- caused by melt convection, interdiffusion, and intermetal-
ing in the transition zone. X-ray diffraction analysis was lic phase formation.
conducted to reveal intermediate phases as shown in Fig. 16.
The peaks belonging to Fe3Al and FeAl2 were visible. The or-
dered iron-aluminum peak is not clearly identified due to the 4. Conclusions
The following conclusions can be drawn:
The fluid dynamic design of wire arc spray nozzles has a
strong influence on the coating characteristics, and coating
properties can be improved by relatively minor modifications of
the spray gun design. Higher gas velocities at the wire tip loca-
tion can be achieved either by shaping the atomizing gas nozzle
such that proper expansion of the flow reduces the formation of
shocks, or by the introduction of a secondary gas in a plenum up-
stream of the wire guides. While use of air as atomizing gas re-
sults in increased oxidation when the particle size is reduced,
using nonoxidizing gases such as nitrogen or carbon dioxide in
combination with a shroud can minimize this oxidation. The
higher gas velocities and the associated smaller particle sizes
and higher particle velocities result in higher density coatings
and improved adhesion of the coatings. This improved adhesion
Fig. 16 X-ray diffraction pattern of interfacial region of sample can be attributed to the formation of metallurgical bonds, which
sprayed with primary/secondary gas atomization have been observed when the particle velocity and temperature

574Volume 8(4) December 1999 Journal of Thermal Spray Technology

at impact are sufficiently high. Under these conditions, failure Coating Properties, Thermal Spray: Practical Solutions for Engi-

Peer Reviewed
occurs within the film rather than at the interface. neering Problems, C.C. Berndt, Ed., ASM International, 1996, p 577-
583
11. E. Pfender, Plasma Jet Behavior and Modeling Associated with the
Acknowledgment
Plasma Spray Process, Thin Solid Films, Vol 238, 1994, p 228-241
This study was supported in part by the NSF through the En- 12. X. Wang, S. Russ, J. Heberlein, W. Gerberich, and E. Pfender, Arc
gineering Research Center for Plasma-Aided Manufacturing, Spray by a Supersonic Atomizing Gas Stream, Proceedings of the 11th
grant No. EEC 8721545. The government has certain rights in International Symposium on Plasma Chemistry (Leicestershire,
this work. England), J. Harry, Ed., Vol 1, 1993, p 133
13. A.W. Schrader, Modern Approaches to Wettability, Theory and Ap-
References plications, Plenum Press, 1992, p 41-45
1. Thermal Spraying: Practice, Theory and Applications, Committee 14. X. Wang, J. Heberlein, E. Pfender, and W. Gerberich, Effect of
on Thermal Spraying, American Welding Society, Miami, FL, 1985, p Shrouded CO2 Gas Atomization on Coating Properties in Wire Arc
3-12 Spray, Thermal Spray Science and Technology Proceedings, C.C.
2. H.-D. Steffens, Z. Babiak, and M. Wewel, Recent Developments in Arc Berndt and S. Sampath, Ed., ASM International, 1995, p 31-37
Spraying, IEEE Trans. Plasma Sci., Vol 18 (No. 6), 1989, p 974 15. N. Birks and G. Meier, Introduction to High Temperature Oxidation
3. M.L. Thorpe, Thermal Spray Applications Grow, Adv. Mater. Proc-
of Metals, Edward Arnold Ltd., London, 1983, p 67
ess., Vol 134 (No. 4), 1988, p 69
16. H. Herman, Plasma Sprayed Coatings, Sci. Am., Vol 45 (No. 7), 1988,
4. H.-D. Steffens, Metallurgical Changes in the Arc Spraying of Steel, Br.
p 112-117
Weld. J., Vol 13 (No. 10), 1966, p 597-605
5. R.W. Fox, Introduction to Fluid Mechanics, John Wiley & Sons, 17. C.C. Berndt, Instrumental Tensile Adhesion Tests on Plasma Sprayed
1973, p 54 Barrier Coatings, J. Mater. Eng., Vol 11 (No. 4), 1989, p 275
6. S.G. Russ, “ Turbulence and Entrainment in Plasma and Heated Jets,” 18. H.-D. Steffens, D. Wielage, and J. Drozak, Interface Phenomena and
Ph.D. thesis, University of Minnesota, 1993 Bonding Mechanism of Thermally Sprayed Metal and Ceramic Com-
7. David R. Marantz and Daniel R. Marantz, State of the Art Arc Spray posites, Surf. Coat. Technol., Vol 45 (No. 3), 1991, p 299
Technology, Thermal Spray Research and Applications, T.F. Ber- 19. C.C. Berndt, The Need for Standardization of Testing Techniques in
necki, Ed., ASM International, 1990, p 113 Thermal Spray Industry, Thermal Spray Research and Applications,
8. K.S. Fessenden and Z. Zurecki, Sliding Wear Resistance of Electric Arc T. Bernecki, Ed., ASM International, 1990, p 325
Sprayed Coatings, Thermal Spray Research and Applications, T.F.
20. C.C. Berndt and R. McPherson, Adhesion of Plasma Sprayed Ceramic
Bernecki, Ed., ASM International, 1990, p 81
Coatings to Metals, Mater. Sci. Res., Vol 14 (No. 3), 1983, p 619
9. X. Wang, D. Zhuang, E. Pfender, J. Heberlein, and W. Gerberich, Effect
of Atomizing Gas Pressure on Coating Properties in Wire Arc Spraying, 21. F.S. Shieu, R. Raj, and S.L. Sass, Control of the Mechanical Properties
Thermal Spray Industrial Applications, C.C. Berndt and S. Sampath, of Metal-Ceramic Interfaces through Interfacial Reactions, Acta Met-
Ed., ASM International, 1994, p 587-592 all. Mater., Vol 38 (No. 11), 1990, p 2215
10. T. Watanabe, X. Wang, J. Heberlein, E. Pfender, and W. Herwig, Volt- 22. R. Raj, A. Pearce, and C.M. Kennefick, Thin Films of Transition Metals
age and Current Fluctuations in Wire Arc Spraying as Indications for on Oxides, Acta Metall. Mater., Vol 38 (No. 11), 1990, p 2215

Journal of Thermal Spray Technology Volume 8(4) December 1999575