Examination of the binary system Al2O3‐ZrO2 by aero acoustic levitation melting

Abstract

Al₂O₃‐ZrO₂ composites exhibit excellent mechanical and high‐temperature properties. The solidification of various hypoeutectic compositions has been studied by means of aero‐acoustic levitation. A high‐speed camera recorded the crystallization, to the correlation of the video stills with the observed microstructures. Solidification takes place by formation of several nuclei and subsequent growth. Nuclei are formed in the supercooled melt, entailing to a fine‐grained, simultaneously solidified structure. The remaining melt between the growing nuclei is heated due to recalescence leading to primary precipitation of zirconia, followed by eutectic solidification. A consistent behavior is presented to explain the observed microstructures. Additionally, samples between 40 and 50 mol% ZrO₂ exhibit lamellar areas, which exceed the initial zirconia composition. The observed microstructure strongly indicates the existence of a liquid miscibility gap.

INTRODUCTION

Due to the high melting point and stability under atmospheric condition, oxide ceramics are indispensable for high-temperature applications. Among the different compositions Al₂O₃-ZrO₂ is particularly interesting due to an excellent abrasive resistance at high temperatures,¹ hardness, high corrosion resistance, and fracture toughness.² Despite this material is highly used, various publications reported contradictions regarding the phase diagram, which is likely caused by the techniques used for characterization.

Due to the various high-temperature applications and the partial synthesis via the melt, it is of particular interest to understand the material's behavior up to the molten state. Older publications report a simple eutectic system with either no solid miscibility³ or solubility of up to 5 mol% Al₂O₃ in ZrO₂.⁴ Jerebtsov et al.⁵ studied the liquidus surface by means of DTA. The hypoeutectic side showed a steep decrease of the liquidus temperature with increasing Al₂O₃ content. Around 1880°C the liquidus curve flattens and reaches the eutectic temperature of 1866°C at 36.5 mol% ZrO₂. Kamaev et al.¹ reinvestigated the system by DTA. They found a miscibility gap at a composition between 40 and 53 mol% ZrO₂ with a monotectic temperature of 1872 ± 7°C. Moreover, they reported that a subregular ionic solution model best approximates the experimental data. Udalov et al.^6,7 also postulated a miscibility gap in the binary system between 35 and 68 wt% ZrO₂. Fabrichnaya et al.⁸ modeled the phase diagram within their work on the ternary system ZrO₂-Y₂O₃-Al₂O₃. They briefly addressed the slope of the liquidus curve, found by Jerebtsov. Instead, they choose a simple eutectic system, considering the solubility of Al₂O₃ in ZrO₂. The data published by Fabrichnaya is commonly used for the modeling of multi component systems.

In comparison to conventional methods, levitation techniques offer decisive advantages for the examination of ceramic melts. The melt is processed without direct contact to a vessel and thus, limitations due to insufficient container material can be omitted. Additionally, melt contamination, which causes problems, becomes impossible. As there is no favored nucleation site in the levitated droplet, homogenous nucleation occurs. Thus, high supercooling is possible, leading to unique solidification structures. Telle et al.⁹ studied the system ZrO₂-SiO₂ by aero-acoustic levitation and observed the liquid phase separation in situ. Moreover, Li et al.^10,11 studied the solidification behavior of eutectic ZrO₂-Al₂O₃ and mullite using aero-acoustic levitation.

Conventional solidification processes are well described in the literature. In contrast to that, publications on the solidification of ceramics under containerless conditions are rare. This work aims explain the solidification of aero acoustically levitated hypo eutectic specimens in the system Al₂O₃-ZrO₂.

EXPERIMENTAL

The following raw materials were used: Al₂O₃ (Baikowski, Malakoff works, Texas, USA; BMA15), >99.99 wt.% Al₂O₃, containing 15 ppm Na₂O, 20 ppm SiO₂ as impurities, and undoped ZrO₂ (Industriekeramik Hochrhein, Germany), >99.9 wt. % containing (ZrO₂ + HfO₂). The as received powders were blended in different quantities in a mortar with ethanol. Table 1 shows the different prepared powder mixtures. Specimens of around 3 mm were formed by heating and melting the powders by a CO₂ laser in a copper mold. Slightly moving the mold led to circular motions of the viscous bead and to a spherical shape.

TABLE 1 Composition of powder mixtures, examined by aero-acoustic levitation

ZrO2 [mol%]	60	50	45	40	38	36
Al2O3 [mol%]	40	50	55	60	62	64

The experiments were performed using an aero-acoustic levitation melter (Physical Property Measurements Inc., Evanston, Illinois, USA) (Figure 1). A detailed description of the experimental setup is given by Nordine et al.¹² Briefly, the ceramic beads are levitated above a Bernoulli nozzle by a nitrogen gas stream. Additionally, three ultrasonic transducer couples stabilize the sample in the horizontal plane. A diode laser position system detects slight deviations from a central position. In order to counteract positional fluctuations, acoustic forces are adjusted by a feedback controller. The samples are heated by two 240-W-CO₂ lasers (SynradInc, Mukilteo, WA, USA). The laser's power output can be controlled independently in 0.5% increments of the maximum power. A pyrometer (Exactus BF 8402, Bayer Catalysts LCC, Rome, Italy) recorded the sample temperature at 1000 Hz. The temperature was calibrated to the melting point of pure Al₂O₃ 2054°C.¹³ The experiments are recorded via a high-speed camera (V 5.2, PhantomVision Research Inc., Wayne, NJ, USA) at a maximum sample rate of 2600 frames per second. A long-distance objective attached to the camera yields full-scale images of the sample.

FIGURE 1. Aero-acoustic levitator in operation; CO2 laser (left); high-speed camera (right); and transducer couples (middle)9

The experimental procedure begins with the gradual heating of the samples. After the sample is melted and the desired temperature is reached, the sample is cooled by switching off the laser. The release of the latent heat is detected by a pyrometer. Subsequently, this process is rerun several times due to the statistical nature of homogeneous nucleation.

The solidified achieved samples were analyzed by scanning electron microscopy and energy dispersive spectroscopy (FESEM Gemini 500, Zeiss, Oberkochen, Germany; EDS detector X-Max80, Oxford Instruments, Abingdon, Oxfordshire, UK). After examination of the surface, cross-sections were prepared and analyzed again. EDS was used to analyze the global and local composition. No significant deviations from the initial composition were found. Videos of the final cooling were correlated with the microstructure examined by SEM.

RESULTS AND DISCUSSION

The molten samples were cooled down within a few seconds, by switching off the laser. Figure 2 shows an exemplary temperature trace of a bead during cooling. The apparent temperature is plotted as a function of time. Temperature traces indicate an undercooling of approximately 200–500 K below the eutectic temperature of 1861°C,¹ before the recalescence. Interestingly, there were no signs of a thermal arrest due to the recalescence. Thus, the release of latent heat was not sufficient to reach the melting point. The nucleation temperature scatters up to 100 K within one composition. As homogenous nucleation is a statistical process, the deviations are expectable. The highest nucleation temperature of 1674°C was found for 60 mol% ZrO₂. In contrast, samples containing 36 mol% showed the lowest nucleation temperature of 1376°C. An influence of the composition could not be proven beyond doubt due to the scatter of the data. Only the nucleation temperature of the samples with 60 mol% ZrO₂ was significantly higher than the other samples. The maximum recalescence temperatures reached were in the range between 1600 and 1700°C.

FIGURE 2. Exemplary cooling temperature trace, recorded via pyrometer: (1) switching of laser, (2) undercooling below eutectic temperature, (3) recalescence, (4) cooling down

The crystallization recorded via high-speed camera is similar among the different compositions. After the formation of the first nucleus, solidification proceeded rapidly. The nucleus grows and induces the formation of additional nuclei (Figure 3(A)). The number of nucleation sites and size varied among the compositions (Figure 3(A) and (B)). Similar experiments were performed by Li et al.,¹¹ who also observed copious nucleation for Al₂O₃-ZrO₂ eutectic melts in levitation experiments. According to them, the shrinkage due to the liquid solid transformation induces mechanical vibration. These facilitate the formation of additional nuclei. When two growing nucleation sites approach each other, the interface appears brighter (Figure 3(A)). Toward the end of solidification, the brightness contrast is even more pronounced (Figure 3(B)). The brightness contrast is related to the temperature and indicates a local temperature increase between the nucleation cells.

FIGURE 3. Solidification of ceramic beads (approx. 3 mm): (A) formation and growth of multiple nucleation colonies (middle and upper part) (45% ZrO2/55% Al2O3) and (B) local temperature increases due to release of latent heat (bright spots) between solidified growth colonies (50% ZrO2/50% Al2O3)

The solidification microstructures of the samples containing 50 and 45 mol% ZrO₂ are particularly suitable for correlating them with the videos. Based on that, the solidification process will be discussed in more detail. Subsequently, unique features of the samples with different compositions will be explained. Figure 4 shows the solidified bead, which is also shown in Figure 3(B). The homogenous light gray areas (exemplary highlighted by a red line) correspond to growth colonies, which originate from different nuclei. A radial growth structure emerges from the protruding sites in the middle of the cells. Therefore, it is reasonable to assume that solidification proceeds from the protruding site. Li et al.¹⁰ found a comparable structure for Al₂O₃-ZrO₂ and came to the same conclusion. In contrast to Li et al.,¹⁰ zirconia crystals are visible in a dark grayish matrix (exemplary highlighted by a yellow line) surrounding the nucleation cells. These interfacial areas are assigned to the bright spots between the nucleation cells, as seen in the high-speed camera images (Figure 3(A)).

FIGURE 4. SEM micrograph of a solidified bead (50% ZrO2/50% Al2O3). Multiple growth colonies originate from the protruding nucleation sites in their center (light gray/red). Interfacial areas (yellow) exhibit zirconia precipitates (white) in a eutectic matrix (dark gray)

The core of the sample might provide determinant information about the crystallization and nuclei formation, so the cross section was polished and analyzed by SEM. The cross section (Figure 5) exhibits comparable microstructural features. Large zirconia precipitates in a lamellar matrix surround areas with a fine-grained radial grown structure. Consequently, it is plausible that fine-grained cells also represent nucleation sites and that nuclei also form within the bead. The outer areas in the lower part of the sample show a slightly finer structure, than in the upper half. It is assumed that the gas flow leads to better heat transfer by forced convection. Thereby small grain size is favored. EDS measurements (not shown here) confirmed that the composition of the fine-grained nucleation cells is equal to the overall composition. Figure 6 shows the microstructure at a higher magnification. The lighter gray areas correspond to the nucleation sites, which exhibit a small grain size <1 μm. While these areas are primarily irregular, lamellar structures were found as well. Large zirconia dendrites (white) of up to 200 μm are embedded in lamellar matrix (dark gray) between the nucleation cells. In the center of the nucleation cells, there is no evidence of primary ZrO₂ precipitation, which would be expected according to the simple eutectic phase diagram. However, the fine-grained, partly lamellar structure indicates a simultaneous growth of ZrO₂ and Al₂O₃. According to the theory of competitive growth,¹⁴ fast growth velocities (related to deep undercooling) can lead to coupled growth for a compositional range unequal the thermodynamic eutectic. Therefore, the theory seems highly suitable to explain the suppressed primary precipitation of ZrO₂, especially in the lamellar structured nucleation cells.

FIGURE 5. Polished cross-section of a crystallized sample (50% ZrO2/50% Al2O3). Growth cells exhibit a fine-grained microstructure (light gray). Zirconia dendrites (white) are surrounded by a eutectic matrix

FIGURE 6. Simultaneous precipitation of ZrO2 and Al2O3 within the growth colonies (circular light gray areas). Interfacial areas exhibit zirconia dendrites (white) in a eutectic matrix (dark gray). Sample composition (50% ZrO2/50% Al2O3)

The interfacial areas between the nucleation cells exhibit large zirconia dendrites in a very fine-grained cellular lamellar matrix. The mean composition of these regions determined by EDS was equal to the overall composition. A slightly coarser microstructure characterizes the lamellar cell boundaries. Lamellar cells close to large zirconia dendrites exhibit zirconia concentrations between 36 and 39 mol% close to the eutectic composition of 35.5 mol%.¹ Some lamellar cells close to the nucleation sites contained higher zirconia amounts (Figure 7 ). The microstructure of the interfacial areas indicates a stepwise precipitation of large zirconia crystals followed by eutectic solidification.

FIGURE 7. Zirconia enriched microstructure resulting from a liquid phase separation. Initial composition (45 mol% ZrO2/55 mol% Al2O3)

TABLE 2 Chemical composition measured by EDS (Figure 7)

Spectrum	ZrO2 [mol%]	Al2O3 [mol%]
1	50.8	49.2
2	52.6	47.4
3	47.8	52.2
4	47.6	52.4

Importantly, alumina-rich samples (36 and 38 mol% ZrO₂) did not contain zirconia dendrites. SEM micrographs show irregular round areas with a slightly radial growth structure are surrounded by lamellar grown regions. The irregular areas represent the nucleation cells. It is assumed that the local temperature increase between the nucleation cells affects the microstructure and favors lamellar growth. Samples containing 38 mol% ZrO₂ exhibited slightly coarsened (d ≈ 3 μm) spherical zirconia particles surrounded by the lamellar eutectic. The coarsening indicates the primary precipitation of ZrO₂. As these specimens are close to the eutectic composition of 35.5 mol%,¹ the amount of primary precipitated zirconia is expected to drastically decrease. It appears that the decrease also leads to a change from dendritic to spherical shape, which might be expected.

However, zirconia-rich samples (60 mol% ZrO₂) showed a more continuous transition from the nucleation site toward the areas with clear primary and eutectic precipitation. In the middle of the nucleation site, the structure is of irregular lamellar type. Toward the boundaries, radial dendrites grow. The radial growth structure in cross section proves that the fine-grained areas are indeed nucleation sites.

Based on these observations, the following theory explains the solidification process and the resulting microstructure. An initial nucleus forms in the supercooled melt. Solidification shrinkage causes vibration, which facilitates the formation of additional nuclei.¹¹ Due to the fast undercooling, the growth velocity is sufficient to suppress dendritic growth, which would be expected according to the simple eutectic phase diagram. Instead, simultaneous precipitation is favored, which could be explained by the theory of competitive growth.¹⁴ As the solidification proceeds rapidly within less than a few seconds in the experiment, the release of latent heat leads to a temperature increase in proximity to the growth front. As a result, the interfacial area between several growing nuclei is heated significantly, which is confirmed by the high-speed camera videos (Figure 3). Especially toward the end of the crystallization, the temperature of the melt is much higher than in the beginning. The temperature increase leads to decreased growth velocity and solidification closer to equilibrium conditions. As a result, large Zirconia dendrites precipitate, while the remaining melt approaches to the eutectic composition. Finally, the liquid solidifies by the eutectic reaction. Li et al.^10,11 observed different kinds of structures during their levitation experiments of ZrO₂-Al₂O₃ eutectic samples. They found an abnormal eutectic structure in the middle of the samples and lamellar one close to the surface. The morphologies were attributed to a thermal gradient within the sample, resulting in different growth conditions.¹¹ As already mentioned, there is an influence of the position on the microstructure. However, the effect of local temperature distribution due to latent heat release seems to have a higher impact on the microstructure.

Samples containing 40–50 mol% ZrO₂ showed clear signs of a liquid miscibility gap. The samples exhibit zirconia-enriched areas where the zirconia content exceeds the initial composition (Figure 7). These areas have a lamellar structure distinguishable from zirconia primary precipitates. The zirconia content varies, which indicates a spinodal segregation in which the composition of the melt changes continuously. The phenomenon was mainly observed at the transition areas from the nucleation cells to the interfacial region. However, areas with increased zirconia content could also be detected within the fine-grained, more regular nucleation cells. As the structure of the nucleation sites was mostly irregular without well-separated cells (Figure 6), it was hardly possible to place appropriate SEM scanning fields. Therefore, clear evidence for liquid phase separation within the nucleation colonies is missing.

The location and structure of the zirconia-enriched areas can be explained as follows. After the laser is switched off, the temperature decreases below the critical temperature. Since the sample solidifies within a few seconds, only local segregation occurs. The rapid growth suppresses the primary precipitation of ZrO₂ and the monotectic reaction. Therefore, the zirconia-rich melt solidifies in a lamellar structure. As already stated, a local temperature increase between the colonies leads to a decrease in growth velocity. Therefore, the solidification proceeds close to equilibrium conditions. The zirconia-rich melt (L1) decomposes according to the monotectic reaction into ZrO₂ and liquid (L2). Indeed, there are small spherical zirconia particles surrounded by eutectic cells that indicate a monotectic reaction.

Despite the clear SEM images, the high-speed camera images show no evidence of melt segregation. In previous work,⁹ the miscibility gap of the system ZrO₂-SiO₂ system was examined by aero-acoustic levitation. The video in these studies confirmed the liquid phase segregation without any doubt. However, the samples were annealed in the phase field of two liquids before switching off the laser. According to Kamaev et al.,¹ the critical temperature in the system Al₂O₃-ZrO₂ is between 1880 and 1890°C and only slightly above the eutectic temperature of 1861°C.¹ Temperature traces confirmed that the initial temperature of the melt exceeded the critical temperature. The solidification occurs only a few seconds after switching off the laser. Therefore, it is assumed that the liquids separate in small ranges. The assumption is supported by the SEM micrographs (Figure 7). However, such small, separated areas are not distinguishable in the videos.

CONCLUSION

Containerless processing enables homogenous nucleation. Consequently, specimens deeply undercool before nucleation. As solidification progresses, additional nuclei form. Local temperature differences strongly influence the resulting microstructure. Nuclei growing in the supercooled melt leading a fine structure without primary precipitates. We assume that the supercooling leads to fast growth rates, which favor simultaneous precipitation of ZrO₂ and Al₂O₃. Latent heat release leads to a local temperature increase between the growing nuclei, which allows dendritic growth in these areas. Thus, the resulting microstructures are explained consistently.

SEM images strongly suggest the existence of a miscibility gap in the liquid phase region. However, the segregated regions are rather small. As a result, the liquid-phase separation could not be observed by the high-speed camera. The melt passes through the phase-field too fast during solidification. Therefore, local segregation is more likely. The melt segregates by a spinodal mechanism. Thus, the boundaries of the miscibility gap cannot be determined with certainty within our study. Further experiments with a controlled cooling process might prove the liquid phase separation. Our results support the observations of various authors, who reported a liquid immiscibility using a wide variety of measurement methods. Thus, the thermodynamic data sets assuming a simple eutectic system should be reconsidered.