**3. Results and Discussions**

*3.1. Phase Composition and Microstructure*

The XRD profiles of the studied A–AT composites are shown in Figure 2.

In the A-10AT(1450) and A-40AT(1450) composites, only peaks corresponding to the Al2O<sup>3</sup> and AT phases were detected without the TiO<sup>2</sup> phase. This demonstrates that the initial titania powders completely reacted with alumina and transformed into AT during fabrication. A previous study [37] investigated the dynamic phase formation in the temperature range of 20–1400 ◦C for the fabrication of A–AT samples by means of neutron diffraction and differential thermal analysis. The results showed that the formation of AT in A–AT samples occurred at temperatures exceeding 1310 ◦C by the reaction sintering of the Al2O<sup>3</sup> and TiO<sup>2</sup> mixture; TiO<sup>2</sup> disappeared at 1370 ◦C. Violini et al. [38] reported that in the two-step reaction sintering of initial powders (alumina and titania), the formation of AT started at 1380 and ended at 1440 ◦C. In our work, the sintering temperature 1450 ◦C ensured the complete formation of AT. However, in the XRD profiles of the A-10AT(1550) and A-40AT(1550) composites, in addition to the main peaks of the Al2O<sup>3</sup> and AT phases, a small peak of TiO<sup>2</sup> (rutile) phase was detected at around 2*θ* = 7.6◦ (*hkl* = 110). This indicated the existence of a TiO<sup>2</sup> (rutile) phase in the 1550 ◦C sintered A–AT composites. The results of quantitative phase analysis demonstrated that the content of TiO<sup>2</sup> was very limited, with values of 0.5 wt.% and 1.2 wt.% in the A-10AT(1550) and A-40AT(1550) composites, respectively.

According to the previous discussion, the complete reaction from initial titania and alumina to AT can be guaranteed with a sintering temperature higher than 1450 ◦C. The detected TiO<sup>2</sup> phase in these 1550 ◦C sintered A–AT composites could be formed due to a partial decomposition of the AT phase during cooling. It is well known that pure synthetized AT is thermally unstable in the temperature range of 800–1280 ◦C. It tends to decompose, through a eutectoid reaction, into α-Al2O<sup>3</sup> and TiO<sup>2</sup> (rutile) during cooling from sintering treatments [13,39]. This brings disadvantage for the material, such as reduced thermal shock resistance. It is well accepted that the decomposition of AT is a nucleation-and-growth process. Experimental evidence has suggested that AT can be thermally stabilized by limiting its grain growth [40,41]. The heat treatment temperature plays an important role in the grain growth progress. In our study, with a smaller AT grain size in 1450 ◦C sintered

A–AT materials, compared with the one in 1550 ◦C sintered samples, the AT phase exhibited increased thermal stability, i.e., without decomposition. *Materials* **2021**, *14*, 7624 8 of 24

> The characteristic microstructures of the investigated A–AT composites are shown in Figure 3.

**Figure 2.** XRD patterns of studied Al2O3–Al2TiO5 bulk ceramic composites. (**a**) The full XRD patterns at the range of 2 =10~80° and (**b**) partial enhancement at the range of 2 *=* 23~30°, as marked in (**a**) with a yellow rectangle. **Figure 2.** XRD patterns of studied Al2O3–Al2TiO<sup>5</sup> bulk ceramic composites. (**a**) The full XRD patterns at the range of 2*θ* = 10~80◦ and (**b**) partial enhancement at the range of 2*θ* = 23~30◦ , as marked in (**a**) with a yellow rectangle.

the Al2O3 and TiO2 mixture; TiO2 disappeared at 1370 °C. Violini et al. [38] reported that

In the A-10AT(1450) and A-40AT(1450) composites, only peaks corresponding to the Al2O3 and AT phases were detected without the TiO2 phase. This demonstrates that the initial titania powders completely reacted with alumina and transformed into AT during fabrication. A previous study [37] investigated the dynamic phase formation in the temperature range of 20–1400 °C for the fabrication of A–AT samples by means of neutron diffraction and differential thermal analysis. The results showed that the formation of AT Figure 3.

posites.

composites, respectively.

in the two-step reaction sintering of initial powders (alumina and titania), the formation of AT started at 1380 and ended at 1440 °C. In our work, the sintering temperature 1450 °C ensured the complete formation of AT. However, in the XRD profiles of the A-10AT(1550) and A-40AT(1550) composites, in addition to the main peaks of the Al2O3 and AT phases, a small peak of TiO2 (rutile) phase was detected at around 2=27.6° (ℎ = 110). This indicated the existence of a TiO2 (rutile) phase in the 1550 °C sintered A–AT composites. The results of quantitative phase analysis demonstrated that the content of TiO2 was very limited, with values of 0.5 wt.% and 1.2 wt.% in the A-10AT(1550) and A-40AT(1550)

According to the previous discussion, the complete reaction from initial titania and alumina to AT can be guaranteed with a sintering temperature higher than 1450 °C. The detected TiO2 phase in these 1550 °C sintered A–AT composites could be formed due to a partial decomposition of the AT phase during cooling. It is well known that pure synthetized AT is thermally unstable in the temperature range of 800–1280 °C. It tends to decompose, through a eutectoid reaction, into α-Al2O3 and TiO2 (rutile) during cooling from sintering treatments [13,39]. This brings disadvantage for the material, such as reduced thermal shock resistance. It is well accepted that the decomposition of AT is a nucleation-andgrowth process. Experimental evidence has suggested that AT can be thermally stabilized by limiting its grain growth [40,41]. The heat treatment temperature plays an important role in the grain growth progress. In our study, with a smaller AT grain size in 1450 °C sintered A–AT materials, compared with the one in 1550 °C sintered samples, the AT

The characteristic microstructures of the investigated A–AT composites are shown in

phase exhibited increased thermal stability, i.e., without decomposition.

**Figure 3.** SEM images of polished and etched surfaces of Al2O3–Al2TiO5 samples and size distribution of Al2O3 and Al2TiO5 particles in each sample: (**a**,**b**) A-10AT(1450); (**c**,**d**) A-10AT(1550); (**e**,**f**) A-40AT(1450); and (**g**,**h**) A-40AT(1550). Yellow arrows indicate locations of AT phases in A-40AT com-**Figure 3.** SEM images of polished and etched surfaces of Al2O3–Al2TiO<sup>5</sup> samples and size distribution of Al2O<sup>3</sup> and Al2TiO<sup>5</sup> particles in each sample: (**a**,**b**) A-10AT(1450); (**c**,**d**) A-10AT(1550); (**e**,**f**) A-40AT(1450); and (**g**,**h**) A-40AT(1550). Yellow arrows indicate locations of AT phases in A-40AT composites.

A dense microstructure was observed in the A-10AT composites (Figure 3a,c) irrespective of the sintering temperature (i.e., 1450 and 1550 °C). In the back-scattered electron images, AT grains and alumina matrix appeared with light and dark gray colors, respec-

and second-phase AT was observed in the A-10AT composites, according to the histogram of grain size distribution in Figure 3b,d. As the AT content increased to 40 vol.%, the grain size of Al2O3 decreased, whereas the AT grain size increased (Table 2). The AT grains in the A-40AT composites exhibited a distinct irregular shape and heterogeneity in size distribution, as indicated by the arrows in Figure 3e,g. Some AT grains clustered together as a submatrix (4–6 µm), where alumina grains were separated and surrounded by AT grains; a similar phenomenon was also reported in another study [38]. The Al2O3 grain size evidently decreased as the AT content increased owing to the inhibiting effect of second-phase AT particles on the Al2O3 matrix grain growth. This phenomenon was also observed in other alumina-based ceramics, such as alumina–zirconia ceramics [22,42]. As the AT content increased, more pores and microcracks were observed in their SEM images.

This corresponds to the reduced density tendency summarized in Table 2.

A dense microstructure was observed in the A-10AT composites (Figure 3a,c) irrespective of the sintering temperature (i.e., 1450 and 1550 ◦C). In the back-scattered electron images, AT grains and alumina matrix appeared with light and dark gray colors, respectively. Round-shaped and slightly elongated-shaped AT grains (1.3–1.6 µm) were homogeneously distributed and mainly located at the triple points and grain boundaries of the alumina matrix. A relatively narrow distribution of grain sizes for both the Al2O<sup>3</sup> matrix and second-phase AT was observed in the A-10AT composites, according to the histogram of grain size distribution in Figure 3b,d. As the AT content increased to 40 vol.%, the grain size of Al2O<sup>3</sup> decreased, whereas the AT grain size increased (Table 2). The AT grains in the A-40AT composites exhibited a distinct irregular shape and heterogeneity in size distribution, as indicated by the arrows in Figure 3e,g. Some AT grains clustered together as a submatrix (4–6 µm), where alumina grains were separated and surrounded by AT grains; a similar phenomenon was also reported in another study [38]. The Al2O<sup>3</sup> grain size evidently decreased as the AT content increased owing to the inhibiting effect of second-phase AT particles on the Al2O<sup>3</sup> matrix grain growth. This phenomenon was also observed in other alumina-based ceramics, such as alumina–zirconia ceramics [22,42]. As the AT content increased, more pores and microcracks were observed in their SEM images. This corresponds to the reduced density tendency summarized in Table 2. *Materials* **2021**, *14*, 7624 11 of 24

**Table 2.** Values of measured density (*ρ*), relative density (*ρ*relative), and average grain size (*G*) of Al2O<sup>3</sup> and Al2TiO<sup>5</sup> , and flexural strength (*σ*<sup>f</sup> ). flexural strength (*σ*f). **Sintering Temperature**  *ρ* **ρrelative** *G* **(μm)** 

**Table 2.** Values of measured density (*ρ*), relative density (*ρ*relative), and average grain size (*G*) of Al2O3 and Al2TiO5, and


Figure 4 exhibits that microcracks were mainly along the grain boundaries between Al2O<sup>3</sup> and AT; some transgranular cracks, acting as bridges between one alumina grain to another, were present in the AT grains. Such microcracks propagation behaviors reflect the complex stresses in A–AT composites during fabrication. Figure 4 exhibits that microcracks were mainly along the grain boundaries between Al2O3 and AT; some transgranular cracks, acting as bridges between one alumina grain to another, were present in the AT grains. Such microcracks propagation behaviors reflect the complex stresses in A–AT composites during fabrication.

**Figure 4.** SEM micrograph showing microcracks in A-40AT samples: (**a**) A-40AT(1450). The aluminum titanate and alumina grains are marked as "AT" and "A", respectively; (**b**) BSE-SEM image of A-40AT(1550). Alumina grains in dark gray, and aluminum titanate in light gray. Red arrows indicate locations of microcracks. minum titanate and alumina grains are marked as "AT" and "A", respectively; (**b**) BSE-SEM image of A-40AT(1550). Alumina grains in dark gray, and aluminum titanate in light gray. Red arrows indicate locations of microcracks.

The effect of sintering temperature was observed as follows. At higher sintering temperatures with the same composition, the studied A–AT composites presented a coarser microstructure associated with the grain growth of Al2O3 and AT. This phenomenon is particularly evident in the A-40AT composites. In these composites, at a higher sintering temperature (i.e., 1550 °C), the excessive grain growth of Al2O3 and AT was promoted (Figure 5). **Figure 4.** SEM micrograph showing microcracks in A-40AT samples: (**a**) A-40AT(1450). The alu-The effect of sintering temperature was observed as follows. At higher sintering temperatures with the same composition, the studied A–AT composites presented a coarser microstructure associated with the grain growth of Al2O<sup>3</sup> and AT. This phenomenon is particularly evident in the A-40AT composites. In these composites, at a higher sintering temperature (i.e., 1550 ◦C), the excessive grain growth of Al2O<sup>3</sup> and AT was promoted after the

after the formation of the AT phase. Owing to the crystallographic anisotropy of thermal

proximately 2 µm) necessary for microcracking [38,43]. Compared with the A-40AT(1450)

These microcracks function as excellent regions for main crack energy absorption and dissipation, leading to remarkable crack attenuation and deflection. Such a mechanism can be confirmed by the fracture surface morphologies of A–AT bulk ceramic composites

composites, more microcracks appeared in the A-40AT(1550) composites.

formation of the AT phase. Owing to the crystallographic anisotropy of thermal expansion in the orthorhombic structure, the grain growth of AT particulates became more abnormal. Microcracks form once the grain size of AT reaches the critical size (approximately 2 µm) necessary for microcracking [38,43]. Compared with the A-40AT(1450) composites, more microcracks appeared in the A-40AT(1550) composites.

These microcracks function as excellent regions for main crack energy absorption and dissipation, leading to remarkable crack attenuation and deflection. Such a mechanism can be confirmed by the fracture surface morphologies of A–AT bulk ceramic composites (Figure 5). *Materials* **2021**, *14*, 7624 12 of 24

**Figure 5.** SEM images of fracture surfaces of Al2O3–Al2TiO5 composites. Arrows indicate Al2TiO5 particles and microcracks. (**a**) A-10AT composites irrespective of sintering temperature and (**b**) A-40AT composites irrespective of sintering temperature. **Figure 5.** SEM images of fracture surfaces of Al2O3–Al2TiO<sup>5</sup> composites. Arrows indicate Al2TiO<sup>5</sup> particles and microcracks. (**a**) A-10AT composites irrespective of sintering temperature and (**b**) A-40AT composites irrespective of sintering temperature.

Without microcracking, the predominant fracture mode of A-10AT composites (Figure 5a) was intergranular fracture. In contrast, with weaker grain boundaries and microcracks in the A-40AT composites (Figure 5b), transgranular fracture appeared in AT particles in addition to the intergranular fracture along weak boundaries between particles. The existence of these microcracks and "weak" grain boundaries is presumed to impart low mechanical strength; however, flaw tolerance, such as high thermal shock resistance and improved thermal stability, improves [44]. This point corresponds well with the results of flexure strength of the studied A–AT composites, as given in Table 2. With the increase of AT content, the flexure strength of A–AT composites remarkably decreased, mainly due to the existence of microcracks. As the sintering temperature increased from 1450 to 1550 °C, the flexure strength slightly reduced, which was related to the coarser microstructure and lower density of the composites. Without microcracking, the predominant fracture mode of A-10AT composites (Figure 5a) was intergranular fracture. In contrast, with weaker grain boundaries and microcracks in the A-40AT composites (Figure 5b), transgranular fracture appeared in AT particles in addition to the intergranular fracture along weak boundaries between particles. The existence of these microcracks and "weak" grain boundaries is presumed to impart low mechanical strength; however, flaw tolerance, such as high thermal shock resistance and improved thermal stability, improves [44]. This point corresponds well with the results of flexure strength of the studied A–AT composites, as given in Table 2. With the increase of AT content, the flexure strength of A–AT composites remarkably decreased, mainly due to the existence of microcracks. As the sintering temperature increased from 1450 to 1550 ◦C, the flexure strength slightly reduced, which was related to the coarser microstructure and lower density of the composites.

Attempts were made to identify the existence of TiO2 in the microstructure of 1550 °C sintered A–AT composites. However, no significant evidence of TiO2 was detected in the SEM micrographs of the A-10AT(1550) and A-40AT(1550) composites, owning to the insufficient content of TiO2. Attempts were made to identify the existence of TiO<sup>2</sup> in the microstructure of 1550 ◦C sintered A–AT composites. However, no significant evidence of TiO<sup>2</sup> was detected in the SEM micrographs of the A-10AT(1550) and A-40AT(1550) composites, owning to the insufficient content of TiO2.

### *3.2. Neutron Diffraction Patterns and Rietveld Analysis 3.2. Neutron Diffraction Patterns and Rietveld Analysis*

No distinct differences were observed among the TOF diffraction patterns of A–AT samples with the same AT contents at various sintering temperatures; the diffraction spectra mainly varied with the AT content. The representative diffraction patterns of the A-10AT(1550) and A-40AT(1550) composites are shown in Figure 6. As the AT content increased, the peak intensities of the Al2O3 matrix decreased; this is in contrast to the increase in AT intensity. In addition, at the same scanning point, the differences between the spectra in the in-plane and normal directions (measured at 2 = ±90°) are insignificant. No distinct differences were observed among the TOF diffraction patterns of A–AT samples with the same AT contents at various sintering temperatures; the diffraction spectra mainly varied with the AT content. The representative diffraction patterns of the A-10AT(1550) and A-40AT(1550) composites are shown in Figure 6. As the AT content increased, the peak intensities of the Al2O<sup>3</sup> matrix decreased; this is in contrast to the increase in AT intensity. In addition, at the same scanning point, the differences between the spectra in the in-plane and normal directions (measured at 2*θ* = ±90◦ ) are insignificant.

**Figure 6.** Representative TOF neutron diffraction patterns analyzed by Rietveld method for (**a**) A-10AT(1550) and (**b**) A-40AT(1550) samples. Investigated reflections are marked above peaks with ★ for Al2O3 and \* for Al2TiO5. **Figure 6.** Representative TOF neutron diffraction patterns analyzed by Rietveld method for (**a**) A-10AT(1550) and (**b**) A-40AT(1550) samples. Investigated reflections are marked above peaks with F for Al2O<sup>3</sup> and \* for Al2TiO<sup>5</sup> .

In Figure 6, raw data collected by the diffraction method (observed) are represented by a blue line, and data derived by Rietveld refinement are denoted by a red line overlapping with the blue (observed) pattern. Unlike the results in the XRD analysis, only peaks of the Al2O3 and AT phases without evidence of the TiO2 phase are observed in the TOF diffraction patterns of A-AT(1550) samples, which is concerned with the limited content of TiO2. Due to the difficulty in identifying TiO2 in the microstructure and neutron diffraction analysis, it can be considered that the effect of TiO2 on residual stresses of the composites would be extremely limited, and thus the crystal structure and stress analysis of TiO2 were not involved in the following results and discussion. The Rietveld refinement of a two-phase model consisting of α-Al2O3 and AT phases was highly satisfactory for all A–AT bulk samples. The positions of individual peaks of α-Al2O3 and AT phases were In Figure 6, raw data collected by the diffraction method (observed) are represented by a blue line, and data derived by Rietveld refinement are denoted by a red line overlapping with the blue (observed) pattern. Unlike the results in the XRD analysis, only peaks of the Al2O<sup>3</sup> and AT phases without evidence of the TiO<sup>2</sup> phase are observed in the TOF diffraction patterns of A-AT(1550) samples, which is concerned with the limited content of TiO2. Due to the difficulty in identifying TiO<sup>2</sup> in the microstructure and neutron diffraction analysis, it can be considered that the effect of TiO<sup>2</sup> on residual stresses of the composites would be extremely limited, and thus the crystal structure and stress analysis of TiO<sup>2</sup> were not involved in the following results and discussion. The Rietveld refinement of a two-phase model consisting of α-Al2O<sup>3</sup> and AT phases was highly satisfactory for all A–AT bulk samples. The positions of individual peaks of α-Al2O<sup>3</sup> and AT phases were identified

with blue and black tick marks at the bottom of the pattern, respectively. The gray line below the pattern represents the difference between the observed and calculated intensities, which were virtually flat in both the A-10AT and A-40AT composites. This indicates the excellent fits in Rietveld refinement that were achieved for all the reference powders and A–AT bulk samples; the weighted residual error, *Rwp*, ranged from 3 to 7%.

The phase composition corresponding to the measured gauge volumes was evaluated by Rietveld refinement; the volume fraction is shown on the upper right of the diffraction pattern fitting window (Figure 6). Virtually constant AT contents were recorded at different scanning positions of the same sample. The calculated AT volume fractions were 9.5 ± 1 and 40 ± 2 vol.% in the A-10AT and A-40AT composites, respectively; these results agreed well with the nominal values.
