Influence of Grain Refiners on the Wettability of Al₂O₃ Substrate by Aluminum Melt

The experimental materials used in this work were CP-Al and master alloys Al-3Ti-1B, Al-5Ti-1B, Al-3Ti-0.15C, together with 99.7 pct pure alumina substrate. The chemical compositions of materials are listed in Table I. The weight of the aluminum samples was 0.037 ± 0.003 g. The average surface roughness (R_a) was 0.7 and 1 µm for the alumina disk in the conventional and improved sessile drop tests, respectively. R_a is the arithmetic average of the absolute values of the roughness profile along the sampling length at the sample surface. It is determined from the deviations about the centerline within the evaluation length. These values were measured by a light microscopy (InfiniteFocus, Alicona, Raaba/Graz, Austria) over 3 mm distance.

Two types of wetting tests, conventional sessile drop test and improved sessile drop test were conducted. An overview of the experiments was listed in Table II.

The conventional sessile drop test was conducted at Norwegian University of Science and Technology (NTNU). The substrates and samples were first cleaned in acetone assisted by ultrasonic vibration for 7 minutes. Then, the substrate together with the Al sample putting on a graphite sample holder was inserted into the furnace (Figure 1) surrounded by graphite heating elements, which are located in a water-cooled vacuum chamber. After the sample was sent into the chamber, the chamber was enclosed and sealed to ensure that a high vacuum can be established in the chamber. A calibrated k-type thermocouple was placed just below the substrate to monitor the temperature. The samples were first heated to 900 °C with a heating rate of 300 °C/min, and then heated up to 1100 °C with a constant rate of 50 °C/min in a high vacuum of 3 × 10⁻⁴ Pa and then held for 1 hour. A high-resolution camera (Sony XCD-SX910CR, Sony, Millersville) was utilized to take images with a frequency of 6 pictures/min. The shape and size evolution of the sample, as well as the wetting behavior, can be directly observed during the experiments owing to the self-illumination of samples.

The improved sessile drop test was conducted at Jilin University. The same substrate and sample cleaning process as in the previous test was also applied. The chamber with the substrate was evacuated to 1 × 10⁻³ Pa and heated up to 1100 °C with a heating rate of 20 °C/min. It was held at 1100 °C for 10 minutes for further cleaning of the substrate surface before the temperature was lowered back to the target temperature, 1000 °C. Such a procedure has been used by other authors before for keeping the high vacuum as well as to remove the substrate surface contaminants.[28,42] Then, the cold metal was delivered through an alumina tube to the substrate surface and then melted immediately, as described in a schematic drawing in Figure 2. The thermocouple used is a W5-type (W-5 pct Ra/W-26 pct Rh) thermocouple. A charge-coupled device (CCD) camera was used to take pictures from the moment that the sample touched the substrate, which is defined as time zero. Extremely fast heating of metal could help to remove the surface oxide layer, and the pre-contact between the metal and substrate was avoided by this method. Laser illumination[43] was applied in this type of wetting tests instead of self-illumination used by the conventional sessile drop method, giving less scattering of the measured contact angles. The drawback of this method is, however, that the sample surface condition cannot be directly observed. More details about the experimental procedure and equipment can be found in the literature.[41]

After wetting experiments, the solidified aluminum droplets together with the substrate were embedded in epoxy resin, cut, and mechanically polished. The microstructure at the cross-section of the droplet and substrate was characterized by a scanning electron microscope (SUPRA 55-VP, Zeiss, Oberkochen, Germany) equipped with an energy-dispersive spectrometer (Octane, EDAX, Mahwah). An accelerating voltage of 15 kV and a working distance of 10 mm were applied in the SEM observation.

The evolution of the contact angle between the droplets of different aluminum alloys and pure alumina substrate is shown in Figure 3, where the heating curve is also depicted. Note that time zero is given as when the furnace heating was started, and the contact angle is measured once the solid metal sample is changed into a spherical droplet. As a general trend, the contact angle decreases with time in the high vacuum of 3 × 10⁻⁴ Pa in this case. The final contact angle on pure alumina is 89 deg for CP-Al, 83 deg for Al-5Ti-1B master alloy, 82 deg for Al-3Ti-1B, and 79 deg for Al-3Ti-0.15C. The master alloys show a relatively lower contact angle than CP-Al. Furthermore, a faster decrease of contact angle can be seen for all master alloys.

Figure 4 compares the dynamic surfaces evolution of droplets on pure alumina. Floating oxide skin can be seen on the surface of all droplets (pointed by arrows) after 6 minutes at approximately 1000 °C (Figures 4(a) through (d)). The oxide skin of master alloy droplets cracks, while it distributes more evenly on the CP-Al sessile drop. Figures 4(e) and (f) show the droplets of different alloys at the temperature of 1100 °C. A clean surface can be observed for the droplets of all master alloys, while some oxide skin patches are still floating on the CP-Al droplet surface.

Generally, a similar exponential decay trend of contact angle was observed during isothermal heating in the improved sessile drop test, as indicated in Figure 5 at 1000 °C. However, samples become spherical immediately after touching the substrate. A sharper descending of the contact angle with time than conventional sessile drop tests can be observed. For example, it takes more than 14 minutes to decrease until contact angle of 110 deg in the conventional method (Figure 3), while only less than 12 minutes in improved method: within approximately 30 seconds for the Al-5Ti-1B and Al-3Ti-1B master alloys, and 8 and 12 minutes for the Al-3Ti-0.15C master alloy and CP-Al, respectively. This is because the cold metal drops on the hot substrate and melts immediately, without forming more continuous oxide skin on the surface of droplets during the heating process. As expected, a slightly higher final contact angle was observed at the lower temperature of 1000 °C with the same holding period; i.e., 99 deg for CP-Al, and 94 deg for Al-3Ti-0.15C, 89 deg for Al-3Ti-1B, and 87 deg for Al-5Ti-1B on alumina, than those at 1100 °C measured with the conventional sessile drop method. Still, CP-Al shows a higher contact angle than the rest of the grain refiner master alloys and a slower deoxidation process at the beginning period as shown in Figure 5. No dynamic surface behavior could be observed with the laser illumination mode in the present improved sessile drop test, especially due to the instant deoxidation process.

As shown in Figure 6, the solidified droplet contains long rod-like Al₃Ti intermetallic phase, Al-Fe-Si intermetallic phase along grain boundaries, and small TiB₂ particles. Interestingly, the lower part and the top part of the droplet show different solidification structures: a large fraction of bright particles distribute in the shape of networks in the lower part, while the top part is relatively cleaner. A careful examination by EDS shows that grain boundaries in the upper part are mainly composed of Al-Fe-Si intermetallic particles (Figure 6(c)). In the lower part, the network structures are composed of agglomerated TiB₂ particles (Figure 6(f)) together with Al-Fe-Si intermetallic phase. This indicates that a TiB₂ particle depleted zone is formed in the upper part after the TiB₂ particles sedimented into the bottom of the droplet. Surprisingly, TiB₂ particles are also found agglomerating along the droplet surface regardless of the height (Figure 6(b)), which is opposite to their sedimentation nature. Figure 6(d) shows the interface between the droplet and the Al₂O₃ substrate. Besides sedimentation of TiB₂ particles at the interface, no reaction layer can be seen. However, the aluminum melt and TiB₂ particles that penetrated the cavity of the substrate can be observed. Figure 6(a) shows the triple line between the droplet and substrate, where an agglomeration of TiB₂ particles can also be observed. Similar microstructures are detected for the solidified droplet of Al-3Ti-1B by both conventional and improved sessile drop tests.

As shown in Figure 7, similar to Al-Ti-B droplet samples, Al-3Ti-0.15C master alloy also shows the sedimentation and depletion zones of TiC particles, long needle shaped Al₃Ti particles and grain boundary Al-Si-Fe intermetallic particles. The black pits in the sample are pores forming during mechanical polishing. As can be seen, TiC particles not only settle down on the surface of the substrate (Figure 7(d)), but also agglomerate and distribute along the oxide skin at the periphery of the droplet (Figures 7(a) and (b)), regardless of the height. In the sedimentation zone at the lower part of the droplet, the fine TiC particles agglomerate along the grain boundaries (Figure 7(f)) forming a network structure. Nearly no TiC particles can be observed along grain boundaries in the depletion zone in the upper part of the droplet, where Al-Fe-Si intermetallic phase precipitates along grain boundaries. Note that the number density of TiC particles is smaller than that of TiB₂ particles.

The wetting tests with both conventional sessile drop (Figure 3) and improved sessile drop (Figure 5) methods show that the contact angles between liquid aluminum containing grain refiner particles (master alloys) and alumina are lower than those of pure Al. In principle, the liquid-solid contact angle can be affected by several factors such as heterogeneity of the substrate surface, temperature, trace elements in the alloy, vacuum, liquid properties (viscosity, surface tension), etc. For the present work, the parameters such as substrate surface roughness, droplet weight, temperature, time, vacuum state are strictly controlled. This ensures that the change of contact angle is mainly dependent on the alloy parameters. In addition, the evaporation of liquid Al during the wetting experiment is unavoidable and will also have a significant effect on the contact angle evolution.

High vacuum and high temperatures are vital to get fresh metal-solid contact during the wetting test. However, the contact angle can be underestimated due to the evaporation of liquid Al in vacuum.[44] In principle, the droplet height (H) will reduce while the base diameter (D) of the droplet will either stay constant or reduce for pure evaporation process (no wetting involved). In the present case, metal droplet spreads while the height reduces (Figure 8). This indicates a wetting process rather than a pure evaporation process. The master alloys have larger D/H values than CP-Al. This indicates a better wetting of master alloys to the substrate. Meanwhile, the D/H values of Al-Ti-B samples are larger than Al-Ti-C sample. The diameters increase with time and are almost stabilized after 1 hour holding time, while the height decreases and is almost stabilized at the end for all samples. This indicates that the dynamic wetting properties stabilize as well, where the final contact angle is representative for the current system.

For a reactive system, a reaction layer will form at the liquid-solid interface. This would increase the wetting due to the alliance of the new layer to both liquid and solid. For example, Al₄C₃ layer forms for the Al-SiC system[20] and TiC layer forms at the interface between Al-3Ti melt and SiC substrate.[45] To check if any reaction layer has formed at the interface, the solidified droplet was first mechanically detached from the substrate and then the retained aluminum was etched off from the substrate by 10 pct NaOH solution. The interface structure is shown in Figure 9. As can be seen, a large fraction of grain refiner particles, TiB₂ and TiC, exist on the substrate. No reaction layer could be detected on the grain refiner particle-free areas in the substrate surface. This means that the lower contact angle of master alloys is not caused by the reaction layer formation for the present work. This is because most of the Ti content in the master alloys is occupied by the grain refiner particles. Thus, the concentration of free Ti atoms in the aluminum liquid is very low.

The viscosity, which is affected by alloy elements and intermetallic phases, influences the spreading of the liquid.[46] Higher viscosity usually causes a higher contact angle due to the slower spreading rate. Smaller droplet is also favorable for fast droplet spreading.[47] For the present work, the weights of samples were strictly controlled as 0.037 ± 0.003 g, which gives equal droplet size. It was reported that adding Al-5Ti-1B master alloy will increase the viscosity of Al melt at 720 °C.[48] It is expected that the grain refiner particles have the same effects at higher temperatures of 1000 °C and 1100 °C, therefore, a larger contact angle should be observed for the master alloy, which is not the case for the present work. Thus, the influence of viscosity on the wetting at such high temperatures for Al is probably neglectable in our case, and the reduction in contact angle for the master alloys is not supposed to be related to the viscosity change.

In the solidified master alloy droplets, there are a fraction of large Al₃Ti particles, in addition to the TiB₂ particles in Al-Ti-B, and TiC particles in Al-Ti-C. The surplus Ti contents in the alloys (more than necessary to form TiB₂ and TiC particles) reacts with Al, forming Al₃Ti. The TiB₂ and TiC particles are very stable at high temperatures and can survive at the test temperatures in the present work. On the other hand, the Al₃Ti intermetallic phase cannot. Based on the Al-Ti phase diagram calculated by Thermo-Calc®, the Al₃Ti phase does not exist at 1100 °C for any of the used grain refiner master alloys, while it can exist in Al-5Ti-1B and Al-3Ti-0.15C alloy at 1000 °C. This indicates that the Al₃Ti phase found in the sample after conventional sessile drop test at 1100 °C should be formed during the solidification process and this phase should not be responsible for the contact angle reduction.

During wetting test, an equilibrium state will be achieved among liquid droplet, solid substrate, and the surrounding vacuum (with an extremely low pressure of gas atmosphere) as shown in Figure 10. This relationship can be expressed by Young’s equation:where γ is the interfacial tension. Subscripts s, l, v refer to the solid, liquid, and vapor phases, and θ is the contact angle.

For a given value of γ_lv, the contact angle decreases when the γ_sl decreases or γ_sv increases, which gives improved wettability between the solid and the liquid phase.

The high contact angle between Al and alumina in literature[29‐38] indicates a large γ_sl. However, the γ_sl between aluminum and TiC or TiB₂ is much smaller as inferred from the good wettability.[27,28,49] Since the sedimentation of TiB₂ and TiC particles at the bottom of the droplets, which covers a large fraction of the interface between the droplet and substrate (Figures 6 and 7), the interface tension γ_sl between Al and alumina is partly changed to the γ_sl between grain refiner particles and Al. This results in a decrease in the solid-liquid interfacial energy which gives a lower contact angle. In addition, the particles agglomerating at the droplet surface also plays an important role in the contact angle reduction. The interface tension between liquid and vapor, γ_lv, influences the height of the droplet. Since the density of TiB₂ and TiC particles are higher than that of liquid aluminum, an agglomeration of the particles at the surface of the droplet will exert a dragging force to reduce the height of the droplet. As a consequence, the effective γ_lv in Eq. [1] will be reduced, which will cause a further decrease in contact angle. For the conventional sessile drop test, the Al-Ti-C and Al-Ti-B master alloys show a very similar contact angle to alumina substrate. However, with higher accuracy, the improved sessile drop test shows that the Al-Ti-B master alloys have a lower contact angle than Al-Ti-C. This may be attributed to the difference in the volume fraction of grain refiner particles in the two different types of master alloys. The TiB₂ and TiC contained in the master alloys are corresponding 3.2 wt pct, and 0.07 wt pct for Al-Ti-B (both Al-3Ti-1B and Al-5Ti-1B) and Al-3Ti-0.15C master alloys, respectively. Also, the density of the TiC particles settled on the substrate surface is lower than that of TiB₂ particles (Figures 6(d) and 7(d)). Furthermore, the size of TiB₂ particles is much larger than TiC, which may also increases the drag force and caused a lower contact angle.

The removal of oxide skin at the surface of the droplets is believed to have a major influence on contact angle evolution. The measured contact angle between Al on Al₂O₃ in literature lies between 67 to 110 and 60 to 97 deg[30‐34] at 1000 °C and 1100 °C, respectively. This large scattering of the measured contact angle data is mainly ascribed to the existence of oxide skin on the droplets, particularly at relatively lower temperatures (T < 1000 °C).[32] The original thickness of the oxide skin can be affected by environmental conditions, sample purity, pre-treatment, vacuum status, etc. In an Al-Al₂O₃ system, the contact angle decreases with the thickness reduction of this oxide skin layer. The removal of oxide layer can be evaluated by Reaction [2] with fast heating.[31] At 1000 °C, the critical equilibrium partial pressure of Al₂O according to Reaction [2] is 4.3 Pa, under which, the oxidation layer on Al surface will be removed. For the present study, both temperature and vacuum state fulfilled the requirement for such a reaction to happen.

For the present case, the master alloys were observed to have a faster deoxidation process than CP-Al in all cases, as illustrated in Figures 4 and 5. This is ascribed to the grain refiner particles distributing along the droplet surface. Since the grain refiner particles are heavier than the Al melt, sedimentation of the particles happens during the wetting test, which resulted in the particle depletion zone in the upper part of the droplets. However, the particles at the periphery of the droplet kept staying at the surface, implying that these particles have been trapped by the surface oxide layer of the droplet. This indicates a strong adherence exists between the surface aluminum oxide film and the grain refiner particles. Meanwhile, the particles agglomerated along grain boundaries in the particle sedimentation zone is a result of solidification frontline accumulation.[50] The adherence of grain refiner particles on the oxide films may have also the influence on enhancing the rapture of the oxide skin at the beginning of wetting test, due to the dragging force of the particles. In the late stage of wetting test, the existence of a dense population of grain refiner particles on the oxide skin will slow down the re-oxidation by hindering the transportation of fresh Al atoms to the surface.

Figure 11 presents a schematic drawing to show the behavior of grain refiner particles in Al droplets during wetting experiments. In the beginning, Al droplet is covered with relatively thick oxide skin, most of which is from the original oxide layer at the surface of solid sample. Inside the droplet, the floating grain refiner particles move around due to the internal convection of liquid. Some of the particles move to the surface oxide skin and being captured (Figure 11(a)). During heating, rupture of the thick oxide layer will happen due to the incompatible expansion between liquid and oxide layer. The grain refiner particles adhered to the oxide layer can help the oxide rupture process by impingement (Figure 11(b)). Once the thick oxide layer is ruptured, the liquid metal can spread more freely, and cause a sharp reduction in contact angle (Figure 11(c)). This spreading may become faster due to the downward dragging force of grain refiner particles on the oxide skin. It is important to mention that the Al melt will still reoxidize[39] in the high vacuum in the present wetting experiments, even though the pumping system would take away the Al₂O gas continuously until a dynamic equilibrium between re-oxidization and deoxidization is achieved. The grain refiner particles along the oxide skin will also hinder the transportation of Al atoms to the surface. As a result, the balanced thickness of the oxide layer should be thinner for master alloys than CP-Al. Thereby, a better wetting is expected. In the end, a fraction of grain refiner particles have adhered to the oxide skin, and the rest of the particles agglomerate and settle down onto the substrate surface (Figure 11(d)). With a large fraction of grain refiner particles distributed on the substrate, the wetting system has changed from Al—alumina to Al—grain refiner particles—alumina. The agglomeration of grain refiner particles at the triple line (Figure 6(a)) also helped to improve the wettability between the melt and substrate (lower the contact angle).

It has to be mentioned that master alloys were directly used for wetting tests in the current work, while the addition level of grain refiner master alloys is usually 0.5 to 1 kg/ton in cast house. The total number of the grain refiner particles in the master alloy is about three orders of magnitude higher than in the commercial alloys inoculated by master alloys. Thus, the influence of grain refiner particles on the contact angle between commercial Al melt and alumina substrate can be expected to be much smaller. On the other hand, the lower contact angle between the melts of master alloys and alumina is mainly due to the adherence of grain refiner particles to the oxide skin of droplets. However, during the real filtration process, the aluminum melt is continuous while free droplet surface with oxide skins may not exist. In the flowing melt, the sedimentation of grain refiner particles at the surface of porous wall filter may also be difficult. It is rather safe to propose that the influences of grain refiner particles on the wetting behavior filter to aluminum melt might be negligible in the industrial cases. Therefore, the reduced filtration efficiency by grain refiner particles should not be ascribed to the wetting angle reduction. However, the finding of the strong adherence between grain refiner particles and aluminum oxide film in this work may help to explore the reason why grain refiner particles reduce the filtration efficiency, because oxide film is one of the main sources of inclusion in aluminum melt. Research on the interaction between aluminum oxide films and grain refiner particles in aluminum melt during filtration is ongoing and will be reported in future work.