Splat transitional behaviour as a function of substrate preheating
This chapter investigates splat morphology changes on the substrates undergoing different preheating temperatures. Splat transition behaviour, characterised by the parameter of transition temperatures, was demonstrated for different powder/substrate systems, i.e. nickel-series powders (Ni, Ni5Al and Ni20Cr) on stainless steel substrates, nickel-series powders (Ni and Ni20Cr) on chromium substrates and copper-series powders (Cu and Cu9.5Al1Fe) on stainless steel substrates. In the process of substrate heating, the substrate surface property changes including surface chemistry and surface roughness as a function of preheating temperatures are carefully analysed to explore their influence on splat formation and the splat morphology changes.
Substrate surface chemistry measurements
The substrate surface chemistry may be modified during substrate preheating. As discussed in Chapter 2, the evaporation of surface condensates/adsorbates during preheating makes a contribution in constraining splat fragmentation and promoting splat formation with regular shapes. Nevertheless, the surface chemistry is modified at the same time. It is the very thin substrate surface layer that firstly interacts with the molten droplets and influences the splat formation. Therefore, it is of great importance to investigate the surface chemistry variation of stainless steel and chromium substrates undergoing different preheating temperatures.
Stainless Steel substrates
The surface composition changes of stainless steel substrates at different preheating temperatures, obtained through XPS wide scans, are summarised in Table 4.1. The amount of carbon present on the substrate surface decreased greatly with the increasing temperature as adventitious carbon was oxidised and released into the environment. At the same time, the surfaces became enriched in Fe and O, and relatively depleted in Cr. At 573 K, the Cr element disappeared from the surface within the range of XPS depth resolution (around 10 nm). With increasing substrate preheating temperatures, the substrate became covered by the iron oxides.
Meanwhile, the high-resolution spectrum of O 1s was analysed to characterise the evolution of relevant oxygen chemical states as a function of the preheating history. In order to investigate the change of surface moisture with preheating temperature, the narrow-scan oxygen profile was resolved into three binding energies—530.2±0.2 eV, 531.9±0.2 eV and 533.2±0.2 eV, which represent oxide, hydroxide and physically adsorbed moisture . Fig. 4.1 shows the peak fitting results for oxygen (Fig. 4.1a) and the atomic concentration for each compound (Fig. 4.1b). The parameters of binding energies and FWHM used in the present work are summarised in Table 4.2.
The results of the oxygen peak fitting and the relative atom percent of oxide, hydroxide and moisture can be seen in Fig. 4.1. The ratios of hydroxide/oxide and moisture/oxide decreased dramatically as the substrate preheating temperatures increased. At elevated temperatures, the metal oxide component increased due to thermal oxidation, while the hydroxides (chemically adsorbed) and moisture (physically adsorbed) decreased. In general, water adsorption on the substrate surface, depending on the substrate temperature, substrate materials, the intrinsic reactivity of the surface and the surface defect sites, is generally classified into two main forms: physisorption of molecular water and the chemisorption of molecular water . The physisorption of molecular water corresponds to the very weak interaction/adhesion moisture on the substrate surface, while the chemisorption of molecular water represents the hydroxides with much stronger adhesion. For substrates heated to 423 K, the moisture was effectively removed. Notably, reducing the chamber pressure in the vacuum plasma spraying process can remove surface moisture/adsorbate at low pressure [1, 4, 115] besides substrate preheating. Even at 573 K, a certain degree of hydroxides were thermally stable on the substrate surface; they could dehydrate into gas vapour during hot droplet spreading and be the source for the pore formation. It should be noted that the surface composition changes when the stainless steel substrates undergo thermal treatment to remove surface moisture as shown in Table 4.1. This thesis focused on understanding the change of surface moisture and hydroxides as a function of substrate preheating temperatures. The results of Cr2p3/2 and Fe2p3/2 peak fitting are presented in Appendix A. At room temperature, the compounds of Fe and Cr both showed the presence of metal and a complex mixture of oxides. Under the condition of low preheating temperatures (below 473 K), the substrate surface consisted of both chromium hydroxide and iron hydroxide. With preheating temperature increasing to 473 K, the chromium hydroxides disappeared. In addition, no metallic iron or chromium species were present on the substrate surface above 473 K. At 573 K, the substrate surface was enriched in iron and totally covered by iron oxide and iron hydroxide.
The surface composition changes of chromium substrates at different preheating temperatures, obtained through XPS wide scans, are summarised in Table 4.3. The impurity element Pb was introduced into the chromium substrate during manufacture (according to the composition information from the supplier). The amount of carbon present on the substrate surface decreased because this adventitious carbon burns off as the temperature is raised. At the same time, the surfaces became oxidised, forming chromium oxides.
The high-resolution spectra profile of O 1s was analysed to characterise the evolution of relevant chemical compounds as a function of the preheating history for chromium substrates. Fig. 4.2 shows the peak fitting results for each element (Fig. 4.2a) and the atomic concentration for each compound (Fig. 4.2b). The parameters of binding energies and FWHM used in the present work are summarised in Table 4.4. The ratios of hydroxide/oxide and moisture/oxide decreased as the substrate preheating temperatures increased. At elevated temperatures, the metal oxide component increased due to thermal oxidation while the hydroxides (chemically adsorbed) and moisture (physically adsorbed) decreased. For substrates heated to 423 K, there was still a certain fraction of moisture (~8 %) remaining on the substrate surface.
Investigation of substrate surface roughness
The surface roughness of mirror-polished 304 stainless steel (SS) and chromium substrates, under different preheating conditions, was measured by SPM and 3D optical microscopy techniques. The scanning area sizes for SPM and 3D optical microscopy techniques were 50 x 50 µm2 and 480 x 360 µm2 respectively. The larger scanning area size generally represents the area covered by a single splat in this study, showing a more average assessment of the surface topography. It has been found that the substrate surface roughness generally influences droplet wettability and droplet cooling rate (due to a large contact area), but may not be dominant in changing splat formation if the surface roughness is in nanoscale, for example Ra 200 nm [7, 119, 130].
304 stainless steel substrates
Fig. 4.3 shows the surface topographic information from the stainless steel substrates preheated at different temperatures obtained through SPM technique and Table 4.5 gives the substrate surface roughness measurements. With increasing preheating temperature, more oxide peaks were formed on substrate surface. Overall, more oxide peaks were formed on substrate surface with increasing preheating temperature, but all parameters of Ra , Rq and Sk did not change significantly in nanometre-scale.
The substrate surface changes with different preheating temperatures showed similar trends through using the different analysis techniques of SPM and 3D optical microscopy. Even though the measured values from the 3D optical microscopy were relatively larger than those of the SPM probably due to the different mechanisms in calculating surface heights, both methodologies were acceptable. In general, with increasing temperatures, the substrate surface became rougher and more peaks were observed on the surface. It should be noted that once the stainless steel substrates were heated to 573 K, the parameter of Sk became positive and may play a role in affecting splat formation through modifying droplet wetting behaviour.
This phenomenon will be further discussed in Chapter 5.
The 3D surface topology of chromium substrates under different preheating conditions is shown in Fig. 4.5 and the surface roughness measurements are summarised in Table 4.7. It can be seen that the substrate surface roughness was kept consistent below the preheating temperature of 573 K. Therefore, the effects of parameters induced by surface roughness changes (such as wettability and interfacial friction and heat transfer) on droplet spreading are expected to be minimal.
In summary, more oxide peaks were observed on the substrate surface preheated at higher temperatures. The substrate surface of both stainless steel and chromium materials became slightly rougher but the surface roughness was still maintained in nanoscale. Nevertheless, the substrate surface topology under these relatively low preheating temperatures was kept consistent, which provides a foundation that the surface roughness would not modify droplet spreading dynamics and splat formation.
Characterisation of splat transition temperatures
It is well established that plasma-sprayed splats undergo a transition from splashed shapes to disk shapes when the substrate is heated above a critical temperature. This critical temperature is defined as the transition temperature, above which more than 50% of the splats are disk shaped. This section aims to interpret the transition temperature parameter by taking into account different splat/substrate material combinations.
Splat transitional behaviour
On substrates preheated at different temperatures, a range of splat morphologies can be observed. Below the transition temperature, splats are extensively fragmented while disk-shaped splats are formed above the transition temperature. In order to characterise in detail the transition behaviour of splats, the observed splat morphologies, as a function of substrate preheating temperature, should be carefully classified. In this work, the splats were categorised into four typical geometries for the Ni-series powders and Cu-series powders on the mirror-polished substrates of stainless steel and chromium, as shown in Fig. 4.6, Fig. 4.7 and Fig. 4.8 respectively. The extent of splat splashing decreased in progression from type 1 to type 4. For type 1 splats, the majority of material broke up and detached from the substrate surface during spreading. Accordingly, only a central core with a surrounding solid ring remained on the substrate. The central core had a smooth surface with some voids. Between the central core and the solid ring, no finger streaks were observed. Type 2 splats showed a lower extent of splashing than type 1, with more material remaining on the substrate. Similar to the type 1 splat, the central core was flat. An obvious solid ring from which the finger tips ejected away and diverged was frequently observed surrounding the central core. Differing from the type 1 shape, more finger stripes surrounding the central core depicted the spreading trajectory of splats. Indeed, only a small fraction of materials remained on the substrate surface due to splat disintegration. At the lower substrate temperatures, highly splashed splats mainly consisted of type 1 and type 2, while on hotter substrates, droplet fragmentation disappeared and splats with regular shapes were formed. Even though the material loss was greatly constrained and the majority of material contacted well with the substrate, these regular-shaped splats showed different morphologies. A proportion of droplet material projected out from the periphery, forming splashed fingers: splats with splashed fingers are defined as type 3. Frequently, voids were accompanied with the formation of the long fingers. Type 4 splats were perfect disk shapes with no material extending from the central core. Generally, the central core of type 1 and 2 splats could contact well with the substrate, while most of the material in type 3 & 4 splats could show good contact with the substrate. The degree of interfacial contact will be studied by FIB and will be discussed in detail in the following sections. As seen in Fig. 4.6d and Fig. 4.8c&d, the rim part experiences curling-up due to droplet recoiling and cooling stress [141, 153]. Some disordered oscillations/surface ripples can be observed at the splat edge, while for the central flat area, no oscillations were formed.
This classification of splat morphologies was consistent with our previous studies . Previous research has highlighted that the interfacial heat transfer and solidification process played important roles in forming these different morphologies [8, 22, 88, 99]. Type 1 splats had small cooling rates because the presence of a gas barrier along the interface obstructed heat transfer from the hot droplet to the substrate. Accordingly, the droplet had high kinetic energy to overspread and disintegrate. At the impact region, the high impact pressure could force the vaporised moisture to form supersaturated gas inside the liquid droplet. Compared to type 1 splats, the improved interfacial contact and faster solidification reduced the fragmentation in type 2 splats. Indeed, the type 2 splat could still overspread on the substrate surface. During spreading, some holes could be formed inside the spreading lamellar, which subsequently ruptured the lamellar. The formed gas cushion along the interface due to moisture/adsorbates evaporation from the substrate surface could rupture the high-velocity spreading liquid sheet. The presence of this underlying air generated large upward stress and friction force, destabilizing the spreading edge and puncturing the thin liquid sheet [88, 94, 180]. In the meanwhile, the presence of small unwetted particles, the surface protrusion or obstacle may also contribute to induce these internal holes [98, 180]. These holes inside the liquid lamellar grew fast in all directions due to the surface tension . Once the two edges from two adjacent holes meet, these borders solidified afterwards and remained on the substrate in the form of finger stripes. With the droplet solidification rate was further increased due to the improved interfacial thermal contact, the consumption of droplet kinetic energy was effectively increased. Therefore, the droplet did not have enough energy for overspreading and the splat fragmentation disappeared. Splats of type 3 and type 4 shapes were observed correspondingly. Compared to type 2, the formation mechanism of the observed fingers was different for the type 3 splat. These long/short fingers were induced by fluid instability, for example, the fast solidification or substrate surface protuberance, and projected away from the disk splat. More detail on this effect can be found in the following chapters. It should be pointed out that the fingers splashed over a short distance for splats on the chromium substrate (as shown in Fig. 4.8c) compared to the stainless steel substrate (as shown in Fig. 4.6c and 4.7c), indicating the smaller kinetic energy of jetted materials. To determine the spreading behaviour of different types of splats, three diameter ratios were measured. The overall flattening ratio f Dr / D0 is defined as the ratio of the overall splat diameter ( Dr as shown in Fig. 4.6) to the original average particle size ( D0 54 µm ). Mostly, for type 2 splats, there was an obvious solid ring (diameter Dr ) surrounding the central core. Even though the diameter of this ring was smaller than that of the finger edge, the finger edge appeared ejecting away from this solid ring and diverging. When the spreading diameter of type 2 splats was measured, this remaining ring was carefully identified and located to avoid random choice. There are some literatures identifying the spreading diameter of type 2 splats from the far edge of the splashed finger [181-183]. In this thesis, we tried to precisely and statistically identify the region where the splashed finger started to jet (solid rings for type 1 and type 2 splats). Even though this region was slightly different the edge of splashed finger, it qualitatively described the larger spreading extent compared with the disk-shaped splats. The central flat area ratio c Dc / Dr is defined as the ratio of the central flat area diameter ( Dc ) to the overall splat diameter ( Dr ). The flattening ratio of the central flat area c Dc / D0 is determined by the ratio of Dc to D0 .
Splat transition temperatures
At low substrate preheating temperatures, generally fragmented splats were formed. As the substrate temperature increases, the regular shaped splats without fragmentation were formed. It should be noted that the estimation of transition temperature is strongly dependent on the definition of a disk splat. In this study, type 3 & 4 splats showed predominantly disk shapes with different splashing degrees, while type 1 and 2 were extensively splashed splats. Therefore, a disk splat could justifiably be defined as either all of type 3 & 4 (definition 1), or type 4 only (definition 2, where splashing was minimal or non-existent). The former is more consistent with the definition used by Fukumoto et al. [4, 115].
Frequently, some small sized splats were observed (as shown in Fig. 4.6c & d). The mechanism accounting for the formation of these small splats is postulated to be either particle break-up during flight or secondary droplet ejection upon impact. Once the powders are ejected into the high-temperature and high-velocity plasma stream, some irregular particles and small nodules attached onto the powder surfaces (as shown in Fig. 3.1) may experience break-up and form secondary particles. Generally, these molten secondary particles have smaller diameters compared to the primary impact particles. Accordingly, splats with small diameters can be formed on the substrate surface. On the other hand, in the impact region, compression waves propagating radially may create instabilities along the liquid-solid contact line, rupturing the fluid at the edge and ejecting some tiny droplets to heights of a few millimetres above the substrate, at velocities of 15-20 m/s and at angles of 30°-60° [31, 70, 79, 184, 185]. These secondary droplets then solidified into small disk splats upon impact. Nevertheless, the formation mechanism of these small size splats still remains unclear.
It is necessary to determine the cut-off diameter for the primary impacting and spreading splats. The splat formation mechanism is derived from an impacted droplet with original particle size (45~63 µm) rather than from the secondary droplets or splats. Indeed, the primary droplets have different impact conditions (impact velocity and temperature) compared to the secondary droplets. Approximately 500 splats of Ni or Ni20Cr (including the small sized splats) at 473 K were utilised as an example to determine a suitable cut-off diameter representing the primary spreading splats. The distribution of splat diameters is shown in Fig. 4.9. The splats with small spreading diameters, below 60 µm, occupied the majority by number. Above 60 µm, splat diameters showed a bimodal distribution with two possible thresholds, 90 µm or 150µm. In order to confirm the final diameter threshold, one could assume that all the material from an original spherical particle totally transformed into a cylinder disk splat. The mass conservation of this transition can be expressed by Eq. 4.1,
For one original particle with a size of 40 µm, the calculated splat diameter is 146.1 µm with a thickness of 2 µm, which is a typical thickness of splats formed on the heated substrates Therefore for type 3 and 4 splats of Ni and Ni20Cr, only the splats with diameters greater than 145 µm were evaluated. However for type 1 and 2 splats, it was difficult to characterise the maximum spreading diameter and the exact location where the droplet began disintegrating from the final splat morphology. Generally, the diameter 145 µm was used as the cut-off size for all splats to eliminate undersized splats.
1.3 Outline of the thesis
2 Literature review
2.1 Plasma spraying process
2.2 Impact of millimetre-sized droplets at relatively small velocities
2.3 Splat formation mechanism during plasma spraying
2.4 Gaps in current understanding
3 Experimental methodologies
3.1 Plasma spraying process
3.2 Powder preparation
3.3 Rig setup to collect single splats .
3.4 Substrate preparation
3.5 Substrate surface property measurements
3.6 Observation of splat morphology through Optical Microscopy (OM) and Scanning Electron Microscopy (SEM)
3.7 Observation of splat interface through Focused Ion Beam (FIB)
3.8 Observation of splat interface through Transmission Electron Microscopy (TEM) .
4 Splat transitional behaviour as a function of substrate preheating
4.1 Substrate surface chemistry measurements
4.2 Investigation of substrate surface roughness
4.3 Characterisation of splat transition temperatures
5 Understanding the formation of nickel splats
5.1 Transitional change of Ni splat morphologies
5.2 Numerical study of Ni droplet spreading behaviour
5.3 Role of solidification on droplet spreading
5.4 The formation mechanism of splashed finger
6 Effect of active element Cr in Ni powders on splat formation
6.1 Transitional change of Ni20Cr splat morphologies
6.2 Effects of droplet thermophysical properties on splat formation
6.3 Observation of interfacial characteristics
7 Ni and Ni20Cr splat formation on chromium substrates
7.1 Transitional change of splat morphologies
7.2 Effect of substrate thermal conductivity on droplet spreading through simulation
7.3 Observation of interfacial characteristics for Ni splats
7.4 Observation of interfacial characteristics for Ni20Cr splats
8 Conclusions and outlook
8.2 Future outlook
GET THE COMPLETE PROJECT
Understanding the Formation Mechanism of Plasma-sprayed Ni and Ni20Cr Splats through Experimental and Numerical Study