The other key characteristic of the spray formed microstructure is a fine scale porosity that, at best, is 1% by volume and exists primarily in the form of small voids (1m) between grains (Ref 2). In cold regions of the deposit, such as against a cold substrate and at the edges of the deposit, larger amounts of interconnected porosity are usually found (Ref 2, 30, 41). In addition, in hot deposits (high fractions of liquid), spherical pores can be found that are much larger than the grain size of the deposit (Ref 2, 5, 41). These three types of porosity appear to originate from different causes.

Interconnected porosity appears to arise when the chilling effect of a cold substrate causes the fraction of solid at the surface of the deposit to approach 100% (Ref 1). Under these circumstances, true spray deposition does not occur because filling of the pores between solid particles by excess liquid cannot take place. Recent thermal modeling studies (Ref 30) demonstrate clearly this idea (Fig. 20). The absence of a calculated solidification cooling rate before 0.4 of the normalized thickness for the unheated steel substrate and 0.1 of the normalized thickness for the heated substrate, reflect the minimum thicknesses modeled before any retained liquid is predicted to occur on the deposit surface. These distances match closely the experimental thickness of the deposit containing a high density of interconnected pores (Ref 30, 41). The results of this study (Ref 30) confirm the importance of achieving a rapid buildup of a partially molten (incompletely solidified) surface on the deposit, if a spray deposit of low porosity is to be produced. The results also confirm a well-established empirical feature of successful spray forming, namely the need to have a deposit surface that does not rapidly chill the spray. Empirical studies have shown repeatedly that it is not desirable to have too fine a droplet size in the spray because this leads to a rapid buildup of solid in the spray. It is also important to use a thermally insulating substrate, or alternatively, if the substrate is metallic, then it has to be preheated. This requirement is critical in the spray forming of thick coatings on a metallic substrate, such as spraying tool steel onto metallic iron rolls or spraying stainless steel onto mild steel (Ref 11).

It is generally considered, but not completely confirmed, that the fine scale pores at grain boundaries are ubiquitous to spray formed deposits (Ref 2) and arise from trapped gas from the atomization step. Indirect evidence for this comes from the observation that in the deposition of high-temperature alloys such as copper-and nickel-base alloys deposits are either essentially pore-free or become so after light thermomechanical processing. This change only happens if a chemical reaction occurs between the nitrogen (atomization) gas and a reactive nitride forming element in the alloy such as chromium or zirconium in copper (Ref 42) or nickel (Ref 43). Spraying nickel-base alloys with an inert gas such as argon always produces fine scale porosity that is not readily or permanently removed by subsequent processing (Ref 43). Heavy deformation can close up the pores, but if the gas remains unreacted, then the porosity may return on subsequent annealing.

The coarser spherical porosity that is often reported in deposits produced under hot spray conditions with a high fraction of liquid in the spray occurs as spherical pores that are much larger than the grain size (Ref 2, 41). It appears likely that these pores may form by the coalescence of smaller pores in the presence of a high liquid content. This allows more fluid motion, and the growing pores can push the solid particles apart. Irrespective of the mechanism(s) involved, a simple thermodynamic analysis (Ref 41) suggests that growth of the pores will produce a much larger volume fraction of gas. For finely divided pores with a radius of curvature, r, of a few micrometers, the interfacial energy, , and induced pressure, AP, can reduce significantly the volume of trapped gas. For example with copper ( =1.5 J/m2) and r 3 m:

When a number, m, of these small voids of initial radius, ru coalesce, the radius increases to r2 and the pressure falls, giving a larger volume of gas. Use of the ideal gas law (PV = nRT) shows (Ref 37) that:

and that the volume, V1, of the m small pores increases to a volume V2, where:

Thus, extended coalescence giving the large spherical bubbles (m 1) will result in a significant increase in the porosity observed. If, as is often observed, the coarse gas-filled voids are much larger (r2 60 m) than the fine grain boundary voids (r 3 m) (Ref 2), then relaxation of the interfacial induced pressure (Eq 1) will for the same volume of trapped gas increase the porosity 20 times (Eq 2). This analysis gives a moderate overestimate of the apparent increase in porosity following coalescence because the model ignores the effect of atmospheric pressure. The large spherical gas voids described in Ref 2 were produced in a copper-titanium billet sprayed under conditions of /(s) = 0.6, which should allow for more pore coalescence.

Equation 1 also shows how heavy thermomechanical processing can close up the gas pores even with insoluble gases. If deformation fragments the pores, it will reduce the pore size leading to increased pressure and a decrease in gas volume (m 1. in Eq 3). However, unless the gas in the very small pores reacts chemically, it is always capable of increasing its volume fraction by coalescence during high-temperature solid state processing.

Recently a series of quantitative studies (Ref 37) confirmed fine scale porosity in copper-titanium and the nickel-base superalloy In625. The average porosity in the deposits was measured in copper-titanium billets and In625 tubes. Additional experiments were also performed on copper-titanium tubes and In625 billets. The billets were sprayed onto oxide substrates with a low thermal conductivity. Figure 23 shows the porosities measured in the copper-titanium billets and in the In625 tubes as a function of the modeled fraction of liquid in the spray (Ref 38, 39, 40).

o In

325 tubes €25 billets " jSTi tubes



* In □ c

■ c

j6Ti bi




1 .

^ -I

■ ■

■ *

ï3 G m


0.1 0.2 0.3 0.4 0.5 0.6 0.7 Fraction liquid in spray

0.1 0.2 0.3 0.4 0.5 0.6 0.7 Fraction liquid in spray

Fig. 23 Average porosity in billets and tubes of copper-titanium and In625 as a function of fraction of liquid in the spray. Source: Ref 37, 38, 39, and 40

A steady fall in the level of porosity with increased values of the fraction of liquid in the sprayf(s) can be seen. The two alloys show very similar results, but the behavior is very different in the two geometries. The billets show porosity levels only as high as 8 vol% in very cold sprays f(s) 0.1), falling to 1 vol% asf(s) reached 0.55. In the spray formed tubes, there is a decrease in porosity from a very high value of 25 vol% at af(s) of 0.25 to porosity levels 2 to 3 vol% at high values off(s), such as 0.5 to 0.6. Lavernia and Wu (Ref 5) reviewed the published data on porosity and, as with the results reported here, found high levels of porosity in sprays deposited under cold conditions; the level of porosity decreased as the temperature of the spray increased. Results have also been reported showing the opposite effect. An increase in porosity at high fractions of liquid (Ref 2, 4, 29, and 30), particularly in aluminum alloys (Ref 29, 30) and in alloys sprayed with an inert gas in which there is no interaction between the gas and the alloy constituents has been shown. A minimum in porosity is indicated and present results suggest that this may occur at 0.3 fraction of liquid in the spray in billets (Ref 4). Clearly the optimum value off(s) is higher for tubes than for billets.

The different results between the levels of porosity in billets and tubes show clearly that to obtain equivalent behavior in terms of sticking efficiency and porosity, the spray must be much hotter for tubes than for billets. The difference between these two geometries arises because in billets the surface of deposition remains essentially under the spray at all times. As shown in Fig. 17, the thermal model predicts that the fraction of liquid on the billet surface under the spray rapidly approaches the fraction of liquid in the spray, f(s). The geometry of tube spraying requires that the deposit rotates into and out of the spray with each revolution. Thus, at most, deposition occurs 50% of the time, and the rate of arrival of the spray falls to zero at the "3 and 9 o'clock" orientations. At these positions, high velocity gas is sweeping past the deposit surface, and the mean fraction of liquid on the surface of the deposit,f(d), should be significantly less than that in the spray,f(s). The difference suggests that the control parameter is likely to be the fraction of liquid on the surface of the deposit under the spray, f(d), rather than the fraction of liquid in the spray. Since deposition involves both the spray and the deposit, it is possible that some combined parameter involving both f(d) and f(s) is critical. At present, no reliable three-dimensional thermal models of tube deposition are available. Given such a reliable and validated thermal model, f(d) can be modeled along with the modeled values off(s), and the experimental results on sticking efficiencies and porosity can be correlated with these two parameters. The current results suggest that the thermal conditions in the spray and the deposit are the critical parameters in controlling sticking efficiency and porosity. A wider range of studies is needed, particularly for other geometries such as sheet and plate and with other atomization geometries such as the linear nozzle (Ref 27). The current results do, however, illustrate the value of combining experimental studies with modeling.

0 0

Post a comment