T. Marrocco, D.G. McCartney, P.H. Shipway, Nottingham/UK
A.J. Sturgeon, Cambridge/UK
Paper presented at International Thermal Spray Conference 2-4 May 2005, Basel, Switzerland.
Over the past five years, interest in cold gas dynamic spraying (CGDS) has increased substantially. Considerable effort has been devoted to process development and optimization for low melting point metals such as copper and aluminium. This paper will describe work undertaken to expand the understanding of deposition of titanium by cold spray methods. CGDS deposits have been produced from commercially pure Ti. Using room temperature helium gas, a range of processing conditions, powder size ranges, substrates and substrate preparation methods have been employed to study their impact on deposition of powders. Scanning electron microscopy has been employed to examine deposit microstructures, and microhardness testing of deposits has been conducted. Samples for pull-off bond strength test have been prepared from a number of the more promising sets of parameters and adhesive strengths have been determined. Computational estimates of gas velocity and in-flight particle velocity have been made focusing specifically on the influence that these factors have on the process deposition efficiency. Differences will be discussed in terms of powder feedstock characteristics and the underlying physical and mechanical properties of the powders and substrates.
Cold gas dynamic spray (CGDS) is a coating technology, in which powder particles in the solid state are accelerated to impact with a substrate by a supersonic gas jet. The principle of particle acceleration is well described by the adiabatic expansion of an isentropic 1-dimension flow through a de Laval-type nozzle carrying the powder particles. [1,2] In this way, coating formation is provided via particle impact at high velocity (300 to 1200 m/s) and a temperature much below the melting point of the starting powder (ambient temperature to 700°C) and bonding occurs dueto rapid and significant deformation under high pressures.  It has been reported that many metals such as Cu, Al, Ni, Ti and Ni-based alloys, and even cermets and ceramics can be deposited by cold spraying onto a wide range of substrates. [4,8]
It is generally accepted that the sprayed material must exhibit a certain ductility to allow for shearing of the particle surfaces and subsequent cold welding. A number of previous studies suggested that the particle deposition behaviour is influenced most significantly by the particle velocity prior to impact with the substrate. Particle velocity is a function of the operating conditions including gas type, pressure and temperature, and materials properties, such as particle diameter, density, and morphology. [8,10] For a given material, successful deposition requires a certain minimum particle velocity or 'critical velocity', the value of which depends most significantly on the thermomechanical properties of the spray and also the substrate materials. [10,15] Below the critical velocity, impacting particles would only cause erosion of the substrate. In this sense, it has been proved that deposition efficiency of copper and titanium coatings increases as the gas pressure and temperature increases. 
Some workers have examined the role of substrate surface topography on the formation of bond between incoming particles and substrate. Tokarev  has suggested that particles impacting a substrate in cold spraying first activate the substrate by roughening it; only once this has occurred is a coating able to initiate and grow. It has also been established that with a greater roughening of the substrate surface (going from polished to blasted) deposition efficiency of both copper and titanium increases slightly.  Vicek et al.  have examined the impact of a range of powder types onto a range of substrate materials in cold spraying. They explained differences in the ability of particles to deposit in terms of the mechanical properties of the particles and substrate and the specific impulse of the impact. They have related bonding primarily to the relative ease of deformation of the substrate and particle, and concluded that if the particle was significantly more deformable than the substrate, then adhesion was not possible.
There are relatively few reports addressing in detail the formation of titanium coatings obtained by the cold spray process. It has been shown that deposition efficiency of titanium does not show a strong dependence on the stand off distance in the range of 5 to 25 mm.  In contrast, a denser coating structure has been observed when helium has been used as the accelerating gas (comparing to coatings sprayed with nitrogen) because the particle velocity has greatly increased under the same operating conditions.  This can be explained acknowledging that helium, with higher specific heat and lower molecular weight, produces a higher jet velocity compared to nitrogen. For the same reason, changing the main gas from nitrogen to helium has caused a significant increase in deposition efficiency.  However, Karthikeyan et al.  have also reported the formation of porous titanium coatings by cold spraying even though the deposition efficiency has been larger than 80%.
The present study aims to examine the effect that powder size range, substrate surface condition and main gas stagnation pressure have on bond-strength, porosity, hardness, and microstructure of the titanium deposits.
2. Experimental methods
Two separate batches of commercially pure titanium powder (min 99 wt.%Ti) with different particle size ranges, referred to as CTi (for the coarse powder) and FTi (for the fine powder), were employed in this study as feedstock; both were produced by Hydride-Dehydride process (HDH) and were supplied by Active Metals Ltd., Sheffield, UK. Powder CTi had a nominal particle size range of (-45 to +5 µm); powder FTi had a range of (-25 to +5 µm). The particlemorphology was examined by scanning electron microscopy (SEM) and the size distribution was measured by using laser diffractometry (Laser Mastersizer, Malvern Instruments, UK). To prepare the specimens for SEM analysis the powders were scattered onto an adhesive carbon disk so that all sizes could be observed in their original proportion.
Fig.1. SE image showing CTi powder morphology
Fig.2. Titanium powder size distributions
Coatings were deposited onto Ti6Al4V substrates (45 x 45 x 5 mm) and standard mild steel (45 x 45 x 2 mm) plates in order to analyse the bond-strength and microstructure. The substrate surfaces were finished in different ways, with the specific aim of investigating the impact of surface finish on deposition of the powders. The surfaces were either ground, 1 µm diamond polished or grit-blasted with Al 2 O 3 ; samples were cleaned using methylated alcohol just prior to spraying.
2.2 Cold spraying process
CGDS was performed at the University of Nottingham with an in-house designed de Laval nozzle with an area expansion ratio of ~7.6 and 100 mm long. Using room temperature helium as the carrier and main gas, a range of processing conditions was employed. The main driving gas and the carrier gas were supplied from separate sources, with the carrier gas pressure ~1 bar above that of the main driving gas. The distance between the nozzle exit and the substrate (the stand-off distance) had been previously optimized to 20 mm and this was used throughout the program. A high-pressure powder feeder (Praxair 1264HP, USA) working at a pressure of up to 34 bar, was used to feed powder to the CGDS gun. The powder feed rate (PFR) wheel speed was fixed at 2 rpm. The substrates were placed in a holder for spraying and the CGDS gun was traversed relative to the substrates at a nominal traverse speed of 100 mm/s to generate the coating. Multiple passes of the gun over a given coating were used to increase deposit thickness up to about 900 µm. Trials were carried out in order to study the influence of stagnation pressure, powder particle size range and substrate surface preparation on the final structure of the deposits and their mechanical properties.
2.3 Characterisation of coatings
Microstructural examination of coatings was performed using a scanning electron microscope (SEM, XL-30, Philips) operated at 20 kV. Sections of the coatings were taken perpendicular to the coating-substrate interface in order toobtain cross-sectional images, and they were mounted in conducting resin, ground and polished using colloidal SiO 2 suspension in H 2 O 2 . They were then examined in the SEM using backscattered electron imaging (BSE). Microhardness measurements were made with a Leco M-400 instrument under a 100 gf load and 15 seconds testing time. A series of regularly spaced Vickers indents defined a profile and gave the hardness variation within the samples. The bond-strength of the coatings was tested by using a hydraulic adhesion/tensile equipment (PAT, DFD Instruments, Norway). Coating disks were sprayed onto a larger substrate using a mask (d= 8.16 mm) and subsequently glued to the test elements with a two-component epoxy resin. In order to strengthen the effective bond between the adhesive and the test elements and to reduce glue failure, immediately prior to gluing, the surfaces of the test elements were grit-blasted and the coatings were degreased with methylated alcohol. Porosity levels in coatings were also measured by image analysis from high magnification BSE images of polished cross-sections. A one-dimensional isentropic flow model has been developed to predict gas properties (velocity, temperature, density etc) as a function of position within the nozzle.  From this, the velocity of a particle as a function of position within the nozzle can be computed by an iterative process. Using the particle velocity at the exit of the nozzle (computed as a function of particle size) along with the particle size distribution for a particular powder type, a volumetric velocity distribution was estimated.
Fig.1 shows the secondary electron (SE) image of the as-received CTi powder whilst the size distributions of both powders, as measured by laser diffractometry are shown as cumulative volume percentage plots in Fig.2. From Fig.1 the angular morphology of the titanium powder, as expected for hydride-dehydride processed powders, can be clearly observed. Fig.2 shows that the FTi powder has a sharp cut-off at a lower size limit, which is approximately 7 µm. There is also around 50 vol.% above the nominal 25 µm upper limit. By contrast, the CTi powder has a lowercut-off limit close to 20 µm and there is a substantial volume fraction, ~60%, above the nominal 45 µm upper limit. The mean dimensions are 28 µm and 47 µm for the FTi and CTi powders, respectively.
Fig.3a) BSE image of CTi sprayed onto ground Ti6Al4V at 29bar. Thickness ~530 µm
Fig.3b) BSE image of FTi sprayed onto ground Ti6Al4V at 29bar. Thickness ~990 µm.
Fig.3 shows BSE images of the cross-sections of the coatings sprayed at 29 bar using CTi ( Fig.3a) and FTi ( Fig.3b). The coating deposited from FTi feedstock powder (when compared to the coating deposited from CTi) is significantly thicker (almost twice as thick) but also exhibits a higher degree of porosity that is more regularly distributed. An irregular porous top layer is present in both the coatings to a thickness of ~150 µm.
Fig.4 shows BSE images of the interface of coatings sprayed using FTi powder at two different driving pressures, namely 29 bar ( Fig.4a) and 15 bar ( Fig.4b), which are the extremes of the series of pressures employed. The images show that the impact of the incoming titanium particles has not caused significant deformation of the substrate close to the interface. They also give an indication of the overall porosity, indicating a slightly lower porosity in the case of the coating sprayed at 29 bar. The series of regularly interspaced Vickers indents taken within the cross-sections of all the samples gave a hardness profile for the coatings placed in a band between ~90-180 kgf mm -2 . in comparison, the hardness of the substrate close to the interface was ~340-400 kgf mm -2 .
Fig.4a) BSE image of FTi sprayed onto ground Ti6Al4V at 29 bar
Fig.4b) BSE image of FTi sprayed onto ground Ti6Al4V at 15 bar
Micrographs of the deposits obtained by spraying CTi powder at 29 bar onto Ti6Al4V substrates finished with three different surface conditions are shown in Fig.5. There is no significant difference in the profile at the interface in the cases of polished and ground surfaces, Figs.5a and 5b. In both cases, it is relatively flat and uniform. In contrast, Fig.5c shows an interface with a rougher profile, as expected with grit-blasting. In this case, the incoming particles have naturally deformed to follow the profile of the substrate.
The bond-strength between a coating (deposited with CTi at 29 bar) and substrate is shown in Fig.6 for both Ti6Al4V and steel substrates as a function of substrate surface condition. It can be seen that the grit-blasted surface condition resulted in the lowest bond-strength for both substrate types, with an average strength around 10 MPa, whilst the polished and ground surfaces resulted in higher bond-strengths of ~22 MPa.
Fig.5a) BSE image of CTi sprayed onto polished Ti6Al4V at 29 bar
Fig.5b) BSE image of CTi sprayed onto ground Ti6Al4V at 29 bar
Fig.5c) BSE image of CTi sprayed onto grit-blasted Ti6Al4Vat 29bar
Fig.7 shows the bond-strength of coatings sprayed with FTi powder onto ground Ti6Al4V substrates as a function of the different stagnation pressures employed in the deposition of the coatings. In this case, there is only a small variation in bond-strength within the pressure range, with a monotonic increase in bond-strength with pressure of ~4 MPa. The highest bond-strength is achieved at 29 bar, with the maximum value of ~24 MPa. It is also notable that the standard error in the mean value also increases monotonically with stagnation pressure, indicating a higher variability in the individual results.
Fig.8 shows the results of image analyses to determine the coating porosity. Nine different fields were taken at high magnification and in BSE mode along the polished cross-sections of several coatings. This enabled the porosity of the coatings to be measured. There is no clear trend in coating porosity with stagnation pressure for coatings deposited with FTi powder onto ground Ti6Al4V substrates with the densest structure (~17% porosity) achieved at20 bar. The coating sprayed onto ground Ti6Al4V at 29 bar, using the coarser powder, shows the lowest porosity, with a porosity level of ~13%. Comparing Fig.5 with Fig.4, it is clear that the CTi powder has generated denser coatings than the ones produced by spraying FTi, and this confirms the results obtained by porosity measurements. However, the porosity is generally high for all the coatings with a maximum of ~23 % for the coating deposited from FTi at 25 bar.
Fig.6. Bond strength of CTi sprayed onto different substrates and substrate conditions at 29bar
Fig.7. Bond strength of FTi sprayed onto ground Ti6Al4V at different pressures
Fig.9 shows the cumulative volumetric particle velocity distribution as determined by a one-dimensional model in the cases of FTi sprayed with helium at room temperature and two different pressures, 15 and 29 bar, and CTi sprayed with helium at room temperature and 29 bar. It is clear that the CTi powder generally travels at lower velocity compared to the FTi powder. For example, when a velocity of 600 m s -1 is considered, ~90 vol.% of the CTi powder accelerated by helium at 29 bar pressure travels below that velocity, whereas only ~30% of the FTi powder particles travel below that velocity under the same conditions. Moreover, if the driving pressure is reduced to 15 bar, the fraction of the FTi powder travelling below this velocity increases to ~60%.
Fig.8. Porosity of FTi sprayed onto ground Ti6Al4V at different pressures compared to porosity of CTi sprayed at29bar
The impact of the incoming titanium particles, in both cases of CTi and FTi powders, has not caused significant deformation of the substrate close to the interface, Figs.4 and 5. This can be explained by considering the original hardness of the powders (~170 kgf mm -2 measured with 10 gf load) and the hardness of the Ti6Al4V substrates, which in as-received condition was equal to ~300 kgf mm -2 (measured with 20 kgf load). Clearly, the substrate surfaces exhibit very little deformation, indicating that the deformation associated with the impact will be localized in the incoming particles. This is clearly illustrated in Fig.5c, where it is visible that the incoming particles have deformed to closely follow the rough profile of the grit-blasted substrate.
In general, porosity in cold sprayed material (such as aluminium and copper) is less than 1%. In contrast, the porosity in these coatings is high, irrespective of the spraying conditions and seems to show no clear trend with increasing driving pressure. However, porosity levels of 10 - 30% have been reported in titanium coatings by Karthikeyan et al.  even when deposition efficiency was greater than 90%. These authors attributed the high coating porosity to the porosity within the sponge titanium feedstock powder that they employed. In the current work, the powder particles have low inherent porosity and thus the high porosity in the coating must be attributed to the properties of the particles themselves. This may be associated with the h.c.p. structure of titanium which exhibits very different characteristics under high rates of strain compared to f.c.c. materials (such as copper and aluminium). 
It has been found that the deposition efficiency of the CTi powder at 29 bar is approximately 40%. Assuming that a critical velocity exists above which a particle will deposit and below which it will simply peen the surface, it canbe deduced from Fig.9 that the critical velocity for this powder types is approximately 520 m s -1 . In this case, a deposition efficiency of almost 90% would be expected for the FTi powder under these conditions ( Fig.9). Such a conclusion is supported by the fact that the thickness of the coating deposited from the FTi powder is almost twice that of the coating deposited from the CTi powder ( Fig.3). Such high deposition efficiencies have been reported previously for titanium powders in cold spraying.  Furthermore, it can be seen ( Fig.8) that the porosity in the coating deposited from the CTi powder is significantly less than that of the coating deposited from the FTi powder. Again, Fig.9 indicates that a much larger proportion of the particles in the CTi powder will be travelling below the critical velocity than for the FTi powder, and thus the peening intensity will be higher for the CTi coating, resulting in its low porosity.
Fig.9. Cumulative particle velocity distribution as determined by a one-dimensional numerical model
Fig.6 shows that the grit-blasted surface condition gave the lowest bond-strength for both substrate types, Ti6Al4V and steel, with an average strength around only 10 MPa, whilst the polished and ground surfaces resulted in higher bond-strengths of ~22 MPa. This may be associated with the work-hardening of the grit blasted surface which has even further limited deformation of the substrate, thus hindering the mechanism of surface bonding by gross plasticity. This also implies that in contrast to thermal spraying, the main mechanism of bonding between the substrate and the coating in cold spraying is not mechanical keying but a metallurgical bond. In the application of such coating technology, grit-blasting of the substrate surface is clearly deleterious to the promotion of bond strength. Despite the increase in particle velocity with increase in driving pressure (see Fig.9), Fig.7 shows that the bond-strength of the coatings obtained by spraying FTi powder onto ground Ti6Al4V substrates has only a slight dependence on driving pressure. It is proposed that the main effect of increase in pressure is to increase the proportion of particles travelling at velocities greater than the critical velocity and thus the main effect of increase in pressure concerns the deposition efficiency rather than the bond strength itself.
Titanium has been successfully cold sprayed onto substrates of steel and Ti6Al4V. Whilst deposition was successful, the level of porosity was always relatively high. The bonding was primarily metallurgical, and the bond strength was higher for substrates in the ground condition than it was for grit blasted substrates. It is proposed that deposition characteristics (when compared to those of materials such as copper) are associated with the h.c.p. structure of the titanium.
T Marrocco acknowledges financial support from TWI and EPSRC in the form of a CASE PhD Studentship.
- Dykhuizen R.C. and M.F. Smith: Gas dynamic principles of cold spray, J. Thermal Spray Technol., Vol. 7 (2), (1998), pp. 205/212.
- Van Steenkiste T.H., J.R. Smith, R.E. Teets, et al.: Kinetic spray coatings, Surface and Coatings Technology, Vol. 111, (1999), pp. 62/71.
- Alkimov A.P., A.N. Papyrin and V.F. Kosarev: A method of cold gas-dynamic deposition, Dokl. Akad. Nauk SSSR, Vol. 315 (5), (1990), pp.1062/1065 (in Russian).
- Stoltenhoff, T., H. Kreye, H.J. Richter, et al.: Optimisation of the cold spray process, Thermal Spray 2001: New Surfaces for a New Millennium, C.C. Berndt, K.A. Khor, and E.F. Lugscheider, Ed., May 28-30, 2001 (Singapore), ASM International, OH, (2001), pp. 409/416.
- Li C.-J., W.-Y. Li, Y.-Y. Wang, et al.: Effect of spray angle on deposition characteristics in cold spraying, Thermal Spray 2003: Advancing the Science and Applying the Technology, B. Marple and C. Moreau, Ed., May 5-8, 2003 (Orlando, FL), ASM International, OH, (2003), pp. 91/96.
- Calla E., D.G. McCartney, P.H. Shipway: Deposition of copper by CGDS: an investigation of dependence of microstructure and properties of the deposits on the spraying conditions, Thermal Spray 2004: Advances in Technology and Application, May 10-12, 2004 (Osaka, Japan), ASM International, (2004), pp. 352/357.
- Karthikeyan J., C.M. Kay, J. Lindeman, et al.: Cold spray processing of titanium powder, Thermal Spray: Surface Engineering via Applied Research, C.C. Berndt, Ed., May 8-11, 2000 (Montréal, Québec, Canada), ASM International, (2000), pp. 255/262.
- Zhang D., P.H. Shipway, D.G. McCartney: Particle-substrate interactions in CGDS, Thermal Spray 2003: Advancing the Science and Applying the Technology, B. Marple and C. Moreau, Ed., May 5-8, 2003 (Orlando, FL), ASM International, OH, (2003), pp.45/52.
- Gilmore D.L., R.C. Dykhuizen, R.A. Neiser, et al.: Partilce velocity and deposition efficiency in the cold spray process, J. Thermal Spray Technol. Vol. 8 (4), (1999), pp. 576/582.
- McCune R.C., A.N. Papyrin, J.N. Hall, et al.: An exploration of the CGDS method for several material systems, Advances in Thermal Spray Science & Technology, C.C. Berndt and S. Sampath, Ed., Sept 11-15, 1995 (Houston, TX), ASM International, 1995, p. 1/5.
- Dykhuizen R.C., M. F. Smith, D. L. Gilmore, et al.: Impact of high velocity cold spray particles, J. Thermal Spray Technology, Vol.8(4), (1999), p.559/564.
- Sakaki K., N. Huruhashi, K. Tamaki, et al.: Effect of nozzle geometry on cold spray process, International Thermal Spray Conference, E. Lugscheider and C.C. Berndt, Ed., March 4-6, 2002 (Essen, Germany), DVS Deutscher Verband für Schweißen, (2002), pp. 385/389.
- Vlcek J., L. Gimeno, H. Huber, et al.: A systematic approach to material eligibility for the cold spray process, Thermal Spray 2003: Advancing the Science and Applying the Technology, B. Marple and C. Moreau, Ed., May 5-8, 2003 (Orlando, FL), ASM International, OH, (2003), pp. 37/44.
- Kosarev V.F., S.V. Klinkov, A.P. Alkhimov, et al.: On some aspects of gas dynamics of the cold spray, J. Thermal Spray Technology, Vol.12 (2), (2003), pp. 265/281.
- Garetner F., C. Borchers, T. Stoltenhoff, et al.: Numerical and Microstructural investigations of the bonding mechanisms in cold spraying, Thermal Spray 2003: Advancing the Science and Applying the Technology, B. Marple and C. Moreau, Ed., May 5-8, 2003 (Orlando, FL), ASM International, OH, (2003), pp. 1/8.
- Sakaki K., T. Tajima, H. Li, et al.: Influence of substrate conditions and traverse speed on cold sprayed coatings, Thermal Spray 2004: Advances in Technology and Application, May 10-12, 2004 (Osaka, Japan), ASM International, (2004), pp. 358/362.
- Tokarev A.O.: Structure of aluminium powder coatings prepared by cold gas dynamic spraying, Metal Sci. and Heat Treatment, Vol. 38, (1996), pp. 135/139.
- Vlcek, J., H. Huber, H. Voggenreiter, et al.: Kinetic powder compaction applying the cold spray process - a study on parameters, Thermal Spray 2001: New Surfaces for a New Millennium, C.C. Berndt, K.A. Khor, and E.F. Lugscheider, Ed., May 28-30, 2001 (Singapore), ASM International, OH, (2001), pp. 417/422.
- Thei J.: Modelling of Cold Spray Process, MEng Thesis, University of Nottingham, 2003.
- Hubert W., J.R.Meyer and D.S. Kleponis: Modelling the high strain rate behaviour of titanium undergoing ballistic impact and penetration, International J. of Impact Eng., 26 (2001), pp. 509/521.