On the Role of Bubbles in Metallic Splat Nanopores and Adhesion
- First Online:
- 1.3k Downloads
Highly nanoporous surfaces were observed on the underside of Ni splats. Experiments varying process parameters and substrate treatments were performed to determine the mechanism of pore formation. A theory of impact-induced bubble nucleation and freezing into pores is presented, and calculations are compared with experimental results. Pore formation and morphology is strongly dependent on substrate (a) thermal properties as they affect time for bubble growth before solidification into pores and (b) roughness as submicron scratches enhance nucleation by providing heterogeneous sites and several micron grooves reduce the driving force for nucleation. Splat pull-off experiments are shown that suggest bubble nucleation and pore formation strongly affect adhesion, and represent a strong contribution to the effectiveness of surface roughening. Finally, this observation shows the potential for the manufacturing of high-surface area materials using thermal spray.
Keywordsporosity of coatings roughness effects splats
Splat formation is considered a highly nonequilibrium process, due to the associated rapid cooling (rates have been reported approaching 108 °C/s) (Ref 1, 2), spreading, and solidification (Ref 3, 4). In recent years, the study of single splat morphology and microstructure has greatly expanded, in no small part due to the increasing availability of computational power and high-speed photography. Perhaps most notable is the work by Moreau et al. in which transparent substrates were used to examine splat-substrate interaction in situ at high frame rates (Ref 5). Resolution, however, is limited to optical. In addition, computational modeling of splat impact has steadily improved, incorporating physical properties relevant to droplet spreading and splat solidification (Ref 6, 7). Subsequent morphological (shape) observations have reasonably matched those of experiments and provided insight into the role of nontrivial phenomena such as splat curling (Ref 8), fragmentation, and breakup (Ref 4). Foci of such studies have not only been the modeling of single splat impact and solidification, but also the relation to such actions on overall coating architecture, in particular the wide variety of pores in the coatings (Ref 9). These pores range in size and aspect ratio from highly elongated (and rather ubiquitous) interlamellar ‘cracks’ down to nanoscale (tens to hundreds of nanometer) quasi-spherical pores. The latter have been viewed somewhat in high resolution cross section (Ref 10), and can also be inferred from small angle neutron scattering (SANS) experiments (Ref 11).
These nanopores (in the nanoporosity community, the term ‘micropores’ refers to a far smaller size scale (Ref 12)) represent an important study in the field of thermal spray (TS), especially as the process technology moves from ‘band aid’ to ‘prime reliant’ and more chemically or electronically functional coatings. Their size and potential universal distribution in coatings affect property and transport models in that they cannot be treated as a composite feature, but actually change the properties of the intrinsic material itself (a la nanocomposites (Ref 13)). In addition, when considering the contribution of such a small scale defect to overall material behavior, continuum methods may not be accurate. That is to say, considering the role of a pore in transport behavior, overall conductance C = k(R) * R, where conductivity k(R) is now a function of dimension itself (Ref 14, 15). The high curvature (R ≈ 10-100 nm) makes these pores highly susceptible to sintering and/or coalescence (Ref 16). Finally, their role in intersplat or splat-substrate adhesion is not understood but could be important. Micro(nano)pore formation in splats is currently attributed to fluid breakup and solidification mechanisms in splats, and models have shown that holes of micron-scale can be made via these mechanisms (Ref 9). However, this does not explain the existence of even smaller pores (shown by SANS). In addition, experimental splat characterization efforts have mainly involved top-down or cross-sectional views; few observations have been made of the splat underside, at least not in high resolution. Recently, we showed such observations on Ni splats, revealing a highly nanoporous surface (Ref 17). This was attributed to a high rate of supersaturated gas bubble nucleation at the splat/substrate interface, as a result of initial impact pressure (≈1 GPa) and rapid depressurization within 100 ns. Experimental results compared favorably with analytical models of gas bubble nucleation in the geology literature (Ref 18, 19).
In addition to the surface architecture, one interesting aspect of the Ni splats in the aforementioned study was the relative ease with which they were pulled off the substrate. In TS, adhesion of coatings is paramount; a number of macroscopic tests have been devised to measure pull-off strength (ASTM) and different methods have been proposed to improve this value (Ref 20). One preprocessing technique that has been used for decades is grit blasting, the theory behind which is that the roughening imposed on the substrate increases bond strength via ‘mechanical interlocking.’ Unfortunately, this has never been proven, and although a wealth of experimental evidence has shown that grit-blasting works quite well (i.e., coatings sprayed on flat, polished surfaces do not stick while those on roughened surfaces successfully adhere), few models exist that attempt to account for the large difference in adhesion imparted by the roughness, which is typically of order of up to 5 μm (an exception is work by Liu et al. (Ref 9)). An additional complication is the lack of confidence in typical TS pull-off tests, which are highly dependent upon specimen preparation, and often reflect coating cohesion (Ref 20). Thus, debate still exists as to the origin of improved adhesion, that is to say, does surface roughening affect the mechanical loading on the substrate level, or the intrinsic bonding on the splat level?
In this study, we systematically studied the system of Ni splats on SS substrates, for a number of process conditions, and examined the resulting porosity. We compared experimental observations to bubble nucleation theories in the geology literature, and showed how surface conditions can affect bubble/pore morphology. Finally, we performed a simple comparison of splat-substrate adhesion for different process conditions. Substrates were roughened to different degrees using sandpaper, and relative splat adhesion was assessed using a statistical method (see “Methods” section). At the outset, we hypothesized that the splats on rougher surfaces would exhibit (i) less nanopores on the underside and as a result (ii) improved bonding. The fundamental assertion behind this is that roughness of sufficient magnitude to divert the flow of liquid would significantly reduce the initial impact pressure and depressurization rate, suppressing bubble nucleation at the splat/substrate interface thus providing a greater surface area for splat/substrate interdiffusion. Results supported this hypothesis, and suggest an additional important consideration in the effect of substrate grit blasting in TS.
Materials and Methods
Spray parameters for the experiments
Spray velocity, m/s
Ni particle size, mm
Ar flow, slm
H2 flow, slm
Carrier gas flow, slm
Torch-substrate distance, mm
Powder feed rate, rpm
42 ± 12
42 ± 12
After deposition, optical micrographs (Nikon Epiphot 200) were taken of the substrates, and SEM (LEO1550) and scanning white light interferometry (Zygo 200) were performed on selected splats, to examine morphology and any porosity. Following this, some splats were removed with carbon tape (electron microscopy sciences (EMS)), and the underside examined under SEM. In some cases, splats were ‘folded’ using carbon tape with a slight lateral force, allowing direct comparison of splat and underlying substrate. Adhesion of splats across different substrate conditions was compared in the following manner: Carbon tape was pressed with a 30 kg weight for an estimated average pressure on a 3 cm2 area of the sprayed region for 10 s, and then pulled off. As we hypothesized that rough surfaces would provide greater adhesion, higher pressures were used, that is to say, splats were pulled ‘harder.’ Relative adhesion was assessed by recourse to (a) examination of the tape for areal coverage of splats and (b) comparison of ‘taped’ and ‘untaped’ regions of the substrate. The tape bond strength was of order MPa (Ref 23), so this provided an upper bound for fracture analyses. Pulled-off splats were examined under HRSEM to search for underside bubbles/pores. Image analysis (UTHSCSA ImageTool) was used to quantify pore size and distribution.
Results and Discussion
It is not known to what extent other systems exhibit bubble nucleation upon impact. Reactive metals or materials with low gas solubility could exhibit other gas/liquid responses, affecting nucleation and also splat-substrate adhesion in different ways. Nevertheless, this phenomenon could play a large role in the porosity, contact, solidification, and adhesion in a number of relevant materials.
Although the phenomenon was shown only for Ni, bubble nucleation and nanopore formation could occur in any system with reasonable gas solubility, as long as the gas does not react.
Rough surfaces significantly decrease bubble formation. However, the extent of this decrease has not been quantified. Obviously, impact on pre-existing splats would not lead to such high porosity, but the geometric ‘cut-off’ is not known at this point.
If controlled, this phenomenon provides a potential method for the fabrication of useful nanoporous surfaces using TS.
Although perhaps not the sole mechanism, bubble nucleation likely has an important role in the connection between substrate roughening and TS coating adhesion.
This phenomenon is not fully understood; multi-physics modeling and more experiments are needed.
MQ was supported under NSF grant GOALI-FRG (Ceramics). AG was supported under NSF Career CMS 0449268. V. Srinivasan and J. Colmenares helped with the splat deposition. J. Quinn helped with the SEM imaging. The authors thank H. Herman, S. Sampath, R. Neiser, C. Moreau, H. Zhang, M. Gevelber, and A. Goland for helpful discussions.
- 16.Y.M. Chiang, D.P. Birnie, and W.D. Kingery, Physical Ceramics: Principles for Ceramic Science and Engineering, Har/Disk edn., Wiley, Hoboken, NJ, 1996Google Scholar
- 22.L. Li, A. Gouldstone, and S. Sampath, Intrinsic Properties of Thermal Sprayed Single Splats on Substrate, Thermal Spray Solutions: Advances in Technology and Application (ITSC 2004) (Osaka, Japan), Electronic Proceedings, 2004Google Scholar
- 23.C.K. Shield, C.W. French, and E. Milde, The Effect of Adhesive Type on the Bond of NSM Tape to Concrete, ACI SP230: 7th International Symposium on Fiber-Reinforced Polymer (FRP) Reinforcement for Concrete Structures, American Concrete Institute, November 2005, p 355-372Google Scholar
- 27.A.F. Mills, Basic Heat and Mass Transfer, 2nd ed. Series, Prentice Hall, Upper Saddler River, 1999Google Scholar
- 32.B.C. Earls, Cavitation and Bubble Dynamics, Oxford University Press, Oxford, UK, 1995Google Scholar
- 33.N.H. March, Liquid Metals, Cambridge University Press, Cambridge, UK, 2005Google Scholar
- 36.M. Fukumoto, Y. Huang, and M. Ohwatari, Flattening Mechanism in Thermal Sprayed Particle Impinging on a Flat Surface, Thermal Spray, Meeting the Challenge of the 21st Century, ASM International, 1998, p 401-407Google Scholar
- 37.P. Protopapas, N.A.D. Parlee (1976) Theory of Transport in Liquid Metals Iv Calculation of Interdiffusion Coefficients of Gases in Metals. High Temperature Science 8:141–165Google Scholar
- 38.T. Iida and R.I.L. Guthrie, The Physical Properties of Liquid Metals, Oxford University Press, Oxford, UK, 1993Google Scholar