53 by xiaopangnv


									METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

Cold gas dynamic spray (CGDS) copper deposition: a comparative study of
      single splats on Aluminium alloy (A2014) and steel substrates
     T.Kaireta, G. Di Stefanoa, D.G. McCartneyb, P.Shipwayb, M. Degreza
                 Département de Science des Matériaux et d’Electrochimie ;
                     Université Libre de Bruxelles, Brussels, Belgium
               School of mechanical, materials and manufacturing engineering,
                   University of Nottingham, Nottingham NG7 2AR, UK

A relatively new deposition process called “cold gas dynamic spray” allows the deposition of
coatings by impingement of particles in a solid state. No melting of the sprayed material is
involved. Heat generated on impact, plastic straining and the type of substrate employed plays
a crucial role in the bonding mechanism. In order to improve the understanding of this
mechanism, copper has been deposited on age hardened Al (A2014) alloy and ground flat
stock steel substrates to investigate the role of substrate hardness upon CGDS coating
deposition. By producing single impacts for this study, the effects of one particle interaction
with the substrate is isolated. The work hardening of copper, the heat and the pressure peak on
impact induces changes in the morphology of the particle and the substrate. Depending on its
type and its hardness, different features appear on SEM images. FEM models have shown that
“adiabatic shear bands” (ASB) form at the interface between the particle and the substrate
when good bonding occurs [1, 2]. By comparing the shapes of the impacted particle on the
substrates, the effect of hardness and pressure on the morphology of the splats can presented.
Key words: cold spray, impact, splats, copper coatings
1    Introduction
    Recently, a new thermal spray technology, referred as cold spray, integrated the wide
family of thermal spray technologies (HVOF, plasma, detonation gun arc spray and others).
The process is a high rate coating fabrication technique that involves accelerating fine solid
powders (generally 1-50µm) in a supersonic jet of compressed gas (with velocities ranging
between 500-1000 m/s) on a substrate where upon impact, the particle undergoes severe
plastic deformation and bonds to the surface. A wide range of pure metals, metallic alloys,
polymers and composites can be deposited on to a variety of substrate materials. The gases
used are helium, nitrogen or air and the inlet pressure in the Laval tube reaches 30 bars.
Higher gas velocities can be achieved by preheating the gas up to temperatures of 900K.
    The bonding mechanism is not yet well understood. Assadi and coworkers [2] carried out
extensive computational finite element analysis of single particles impacts. The occurrence of
an “Adiabatic shear instabilities” at the particle to substrate interface and the particle to
particle interface explains the bonding mechanism. The high speed impact induces extensive
localized deformation and the bonding seems to be an entirely solid state process. This
hypothesis is supported by a number of experimental observations:
    a) Ductile materials can be cold sprayed where ceramics can be only cold sprayed if they
       are embedded in a ductile matrix.

    b) The existence of a critical speed shows that the particle needs sufficient kinetic energy
       to plastically deform and/or remove the thin oxide layer to get good bonding. The
       critical speed can be reached if the particle has a minimum size of 5 µm to pass
       through the bow shock. [3]

    c) The particle’s kinetic energy is significantly lower than the one required to melt it on

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

    The deposition of the first layer on the substrate determines the coating’s bonding quality.
The role of the substrate’s properties in the bonding process needs further investigations. The
computational analysis shows that adiabatic shear bands (ASB) occurs either on the substrate
or on the particle depending on the strain rate. The bonding between substrate and the particle
occurs when ASB are present on both particle and substrate. Both of the materials soften and
adopt a viscous behaviour. This phenomenon is called “jetting”. The materials have rolled up
around the rim of the crater. This roll up is the source of the bonding and it is described as
“metallurgical” which suggests that the bonding has a chemical nature.
The formation of the “jetting” is closely related to quality of the bonding of the splat on the
surface. If the substrate does not soften, poor bonding will be observed. Understanding how
substrate properties and impact speed influence the formation of splats will allow better
understanding of the process.
2      Experimental method
2.1 Materials
Commercially available powders from BSA Metal Powders with particle size between 5µm
and 22 µm were sprayed. This is a gas atomised powder. Figure 1 shows a SEM photo of the
powder and the cumulative size distribution of the powder obtained by a laser diffraction




                                                     Cumulative %







                                                                          0.0     10.0   20.0              30.0    40.0       50.0
                                                                                         Particle siz e /microns

    Figure 1 : cumulative size of powder and morphology of the copper powder

Two types of substrate have been chosen for their difference in young modulus and their
difference in hardness. Each substrate has been heat treated in order to get different
hardnesses. Samples have a thickness of 0.505 mm. The first substrate is a rolled sheet of
A2014 and the second substrate is silver steel with the following chemical composition:
Table 1 chemical composition of the steel

       Fe                C               Mn                                 Cr                W                         V
      96.65            0.95%            1.20%                              0.5%              0.5%                     0.20%

Both of these substrates can be heat treated to a wide variety of hardnesses gave values of 100
HV 10kg and 147 HV 10kg with a Vickers indentation on A2014. The hardness of the steel
has been measured with a load of 20 kg and gave 394 HV 20 kg and 866 HV 20 kg.

2.2   Spray deposition
       Cold spraying was carried out at the University of Nottingham using an in house
constructed cold gas dynamic spraying system. The carrier gas and the spray gas used is room
temperature Helium. A De Laval nozzle was used to accelerate the gas to supersonic speeds.

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

The maximum stagnation pressure available was 30 bars and a Praxair 1264HP powder feeder
was employed. The nozzle was fixed vertically to a frame and the sample moved below the
nozzle on a computer controlled X-Y table that allowed faster scans and different shapes to be
deposited. In order to obtain single impacts of particles, the travel speed was set to 0,4 m/s
and the powder feed rate was set to 0,125 g/s. This is the lowest possible setting for the
powder feed rate. The stagnation pressure for the deposition was set successfully from 11
bars, 18 bars, 22 bars, 26 bars to 29 bars.
                            Table 2 : spray parameters used during deposition

                     Stagnation temperature(°C)   22°C
                     Powder feed rate             0,125 g/s
                     Stand off distance           0,02 m
                     Gun travel speed             0,4 m/s
                     Carrier gas                  Helium
      The simple isentropic gas model with a simple model of the particle dynamic allows the
discussion of the basic effects of pressure on the morphology of the splats. Only laser
velocimetry could quantify the value of the speed. The bow shocks will lower the speed on
impact; this effect can be neglected for the particles with a diameter larger then 5 µm [3].
With a stand off distance of 2 cm, the bow shock effect is neglected and the exit velocity is
assumed to be very close to the impact velocity.
3     Results
3.1   General observations
        The formation of single splats depends on the impact speed, the angle of impact, the
size of the particle and the properties of the materials (particle and substrate). The effect of the
substrate hardness and the impact speed (controlled with the gas pressure) are the two
variables in this study. The angle of impact has not been taken in to account.

3.2 Effect of pressure
    The pressure and the size of the particle will determine the impact speed. To keep the
comparison coherent, images of copper particle with approximately the same size will be
compared. These images are compared based on the scale given by the SEM microscope.

                Figure 2 effect of pressure on the A2014 alloy with 100 HV 10kg hardness

Using the isentropic gas flow particle speed, it is possible to estimate the speed difference
between a 10µm particle deposited at 29 bars and a 10 µm particle deposited at 11 bars. This

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

model gives an increase of 200 m/s when deposition is done at 29 bars. This brings an excess
of kinetic energy converted in to heat during impact.

                 Figure 3 effects of pressure on the steel with a hardness of 866 HV 20kg

On A2014 alloy, the substrate has softened in every case. The low melting point of the alloy
compared with copper allows the softening to occur first on the substrate. The deformation
localizes and the heat is generated by the shear stress concentrated on the rim of the impact.
On Figure 2, cross sections of particles show the difference of depth and the importance of the
substrate softening at high impact speed. At low impact speed, the copper particle has kept its
elliptical shape. No sign of copper softening is visible. At 29 bars, the copper particle has
started to soften on the rim of the impact. The heat generated has allowed the copper to reach
a softening point and the cooler top part of the particle has penetrated in the particle.

            Figure 4 impacts of copper particles on steel with a hardness value of 394 HV 20kg

Figure 3 and Figure 4 show copper impact on steel substrates. No significant differences are
visible between a pressure of 18 bars and a pressure of 29 bars. The substrate does not deform
upon impact, the copper particle is considerably softened by the high strain rate. The adiabatic
heating allowed the copper to soften. On the copper impact deposited at 29 bars on the steel
with a hardness of 866 HV 20kg some cracks appear in the soft rim.

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

Figure 5 recristallised structure observed on a crater of A2014 with a hardness value 100HV 10kg deposited at
22 bars.

Figure 5 shows the effect of the high impact speed on the bonding of the particle. The heat is
sufficient to melt a fraction of the substrate. As shown by Barradas and Co [4], liquid state
can be reached and a diffusion layer can form.

3.3 Effect of hardness
Hardness has been modified on each type of substrate by various heat treatments. The
hardness will play a role in the rebound mechanism involved.

                 Figure 6 : crater on steel substrate with a hardness of 394 HV 20 kg at 18 bars

Figure 6 shows crater on the steel with a hardness of 394 HV 20kg. Small impacts have clear
sharp rim. It seems that a very thin amount of substrate material has softened. EDX analysis
has shown the presence of copper on the crater. This is an impact generated by a very small
and fast copper particle. On Figure 7, the craters are left by small copper particle on the steel
substrate with a hardness of 866 HV 20 kg.

            Figure 7 : craters on the steel substrate with a hardness value of 866 HV 20kg at 22 bars

On Figure 7, small copper particle have remained in the rim of the crater. The bonding was
sufficient to allow some parts of the copper particle to remain on the surface. The high

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

hardness value has induced a lot of deformation in some parts of the particles. The heat
generated has been more important since the substrate has not deformed.

Figure 8 : (a) copper particle deposited at 29 bars on a A2014 substrate with 100 HV 10 kg hardness (b) copper
                           particle deposited at 29 bars on a the 147 HV 20kg hardness

On Figure 8, the effect of the hardness is visible by the depth of the impact. At 29 bars, the
impact speed is high and the copper has softened also.

Figure 9 (a) crater on A2014 at P0 =29 bars, jetting comes from the substrate (b) crater on A2014 at P0= 29 bars,
   the rim shows cracks. The presence of intermetallics in the substrates modifies the flow properties of the
                                    materials in some parts of the substrate.

The hardness has also determined the importance of the thermal softening in the A2014
substrate. On Figure 9, craters show the difference between a deposition on a soft substrate
100 HV 10 kg and on a hard substrate 147 HV 10 kg. Figure 9 shows very few signs of
thermal softening and cracks are present in the rim. Heat treatment of the hard substrate has
favored the formation of a hard phase in the alloy rich in Mn and Fe. EDX has shown that this
phase was present on every crack. The absence of extensive jetting on harder substrates
explains weaker bonding on the surface. The substrate needs to have more ductility in order to
promote bonding.
4   Discussion
     In order to understand the bonding of cold spray coating on any substrate, it is a necessity
to determine the fundamental mechanism involved in the bonding. Thermal spray coatings are
bonded by gripping the coating on the surface of the substrate. Grid blasting the surface
allows better bonding by offering more surface. On the other hand, authors [5, 6] describe the
bonding in cold spray as metallurgical which underlies a chemical concept of chemical
bonding. Assadi and Co [3] have showed that the notion of critical speed for bonding of a
particle can be related to the formation of “adiabatic shear band”(ASB). At a high strain rate,
competition between strain hardening and thermal softening determines the possibilities for
shear band to appear. The same authors have shown that melting is not required to get good
bonding; clear conformal and intimate contact between the surfaces and high pressure is
sufficient. When ASB form, the melting temperature has not been reached. Fukanuma and co
[6] have shown that bonding on aluminium reaches value of 40-60 MPa with increasing
stagnation pressure whereas on steel the bonding ranges form 10 MPa to 40 MPa.

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

Understanding the origin of this bonding difference means understanding the effect of the
substrate. The problem of substrate and particle interaction has been approached by
comparing substrates with similar chemical composition but with various hardnesses. The
possibility of melting in a thin layer could provide an explanation for the good metallic
bonding observed ,it has been observed by Barradas and Co [4]. The impact speed of the
particle, due to the helium used, would have provided sufficient energy to melt the A2014
substrate but no sufficient evidence of copper melting has been found.
     Therefore, the speed of the particle determines the quality of the bonding. Using the
isentropic flow model developed by Dykhuyzen [7] it is possible to get a realistic value of the
particle speed a long the centre line of the nozzle. This will allow the discussion to centre
around the effect of speed on splat formation.
                                        speed of particles as a function of size

                                 1200                                              P 11 bars
                   speed (m/s)

                                 1000                                              P 18 bars
                                  800                                              P 22 bars
                                  600                                              P 26 bars
                                  400                                              P 29 bars
                                        1        5        10          15   20
                                                     size (microns)

             Figure 10 : particle speed as a function of size for helium gas at various pressures

4.1    The effect of pressure
        At 29 bars, the particle does not have an elliptical form, the softening does occur. On
Figure 2 (a) and (b), the soft copper shows a lip on the rim of the impact. The kinetic energy
was sufficient to generate heat in a large fraction of the particle’s volume and the lip has
formed. At 11 bars, the impact speed ranges from 600 m.s-1 to 1000 m.s-1 as show on Figure
10 and the copper shows no signs of thermal softening. The impact speed is insufficient to
generate the softening of copper. This suggests that softening of the copper would require
higher impact velocities. The original hardness of the powder could explain the absence of
jetting at low pressure. Harder particles would not reach sufficiently high strain rates and heat
would be evacuated fast enough to prevent the temperature to approach softening conditions.
The heat generated by the high strain rate ( ε =105 s-1) competes with its evacuation by
thermal conductivity. Grujicic and co [1] have shown that with impact speeds beyond 800
m.s-1, the temperature peak reached is close to the melting temperature of aluminium. Melting
can not be excluded. Due to the very short contact time, it is admitted that diffusion does not
take place [1, 2]. Simulations tend to overestimate the temperature reached during shear band
formation; the evacuation heat by thermal conductivity is not taken into account. This means
that melting would occur at temperatures beyond a corresponding impact speed of 800 m.s-1.
At 29 bars, the speed ranges from 1000 m.s-1 for 10 µm particles to 1400 m.s-1 for 1 µm
particles. The rapid solidification process has been observed in the work by McCartney and
co[8] for materials with a low melting point like Tin. Melting is a possibility. Figure 4 and
Figure 3 show splats on steel. No major difference has been shown between 29 bars and 18
bars. The copper particle opens up completely upon impact due to the very high hardness
value of the substrate. Figure 5 shows the effect of the high impact speed on the material. On
Figure 5, the substrate shows clear signs of recristallisation. These cell structures are

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

compatible with the observation made by S. Barradas and Co[4]. They also show the
localisation of the heat generation in the bonding process. Copper has a higher melting
temperature compared with A2014. It is likely that melting occurred on the alloy but not the
particle. The melting temperature of copper is 1360K where the substrate has a melting
temperature of 940K.

4.2    The effect of hardness
        The hardness and the young modulus of the substrate will determine the amount of
elastic energy stored by the substrate [9]. A hard substrate requires a higher impact velocity
to get plastic straining and more energy is available for elastic rebound. Therefore adhesive
energy has to overcome rebound energy in order to keep the particle on the substrate. In CS,
this adhesive force is sufficient to avoid removing the deposited particle by the following
impacts. On steel, the contact surface is considerably enhanced and the particle does not
penetrate in the substrate. The absence of deformation of the substrate excludes the possibility
of shear band formation on the substrate. Some form of thermal softening is visible on soft
steel 394 HV 20kg (Figure 6) for small craters but this is not the case for large particle. There
rim is continuous and EDX analysis shows the presence of copper traces. These very small
impacts have the highest values of impact speed. These craters are not visible on steel with
high hardness (866 HV 20 kg), instead craters show damage on the surface and remains of the
copper particle still remains in the rim (Figure 7). This suggests that some form of good
bonding does exist. The adhesive energy was not sufficient to keep the whole particle on the
substrate but a part of it remained. Figure 4 and Figure 3 show splats on steel with two values
of hardness. On the substrate with a hardness value of 866 HV 20 kg (Figure 3), the copper
jetting has rebounded on the surface. The softened part of the copper is no longer in contact
with the surface. On Figure 4, the copper’s jetting is in contact with the steel’s surface. Splat
formation on aluminium has shown that the copper particle does not deform considerably
compared with splat formation on steel. The jetting comes from the substrates thermal
softening under the high strain rate involved during impact (Figure 2 and Figure 8 and Figure
9). The aluminium has a melting temperature (940K) lower than copper (1360K), thermal
softening will occur first on the substrate. Figure 9 show the importance of the flow properties
of the substrate. The heat treated sample (Figure 9 (b)) shows a number of cracks in the rim.
The aluminium alloy is less ductile and the copper impact does not allow as much substrate
material to soften. Lower ductility would be unfavourable for bonding and a coating would
have weaker bond strength.
5     Conclusions
      • The bonding is improved by higher impact velocities. More heat is generated upon
        impact and the copper particle shows signs of thermal softening.
    • Hardness play an important role if the substrate has not softened. On a ductile
        substrate like aluminium, the high strain rate induced allowed better bonding. Where
        as on steel, the jetting has shown big differences in morphology. Elastic energy is
        more important on very hard substrates.
The use of higher carrier gas pressure is a possibility to improve bonding. The deposition of
an intermediate ductile bonding layer before depositing on hard substrates can also improve
the bonding.
6     Acknowledgements
This work is financed by the Walloon Authorities (DGTRE) and the European union (ESF).
Special thanks to Prof. McCartney for accepting me in his group.

METAL 2005                                    24.-26.5.2005, Hradec nad Moravicí

7   References
[1] M. Grujicic, J.R. Saylor, D.E. Beasly, W.S. DeRosset , D.Helfritch Appl.Surf.Sci 219
(2003) 211
[2]H. Assadi, F.Gärtner, T.Stoltenhoff, H.Kreye, Acta Mater 51 (2003) 4379
[3] B. Jodoin, J.Thermal.Spray.Technol. 11 (2002) 496
Tech.(in press) 2004
[5]M. Grujicic, C.L. Zhao, W.S. DeRosset, D.Helfritch, Materials and design (in press)(2004)
[6] H. Fukanuma, N. Ohno, thermal spray 2004, ITSC 2004
[7] R.C. Dykhuizen, M.F. Smith, D.L. Gilmore, R.A. Neiser, X. Jiang, S.Sampath,
J.Therm.Spray.Technol 8 (1999) 559
[8]D. Zhang, P.H. Shipway, D.G. McCartney, thermal spray 2003 (Ed) C. Moreau and B.
Marples (2003) 45
[9]M.Xu, K.Willeke, J. Aerosol Sci 24 (1993) 19


To top