Modification of YSZ fiber composites by Al2TiO5 fibers for high thermal shock resistance

Yttria-stabilized zirconia (YSZ) fiber composites are highly efficient thermal insulating materials; however, the poor thermal shock resistance limits their versatile applications. In the present study, YSZ fiber was mixed directly with Al2TiO5 fiber, which had an extremely low thermal expansion coefficient, to prepare YSZ−Al2TiO5 (ZAT) fiber composites by compression molding and heat treatment. The minimum thermal expansion coefficient of the prepared ZAT fiber composites was measured to be 7.74×10−6 K−1, which was 26% lower than that of the YSZ fiber composites (10.42×10−6 K−1). It was shown that the prepared ZAT fiber composites maintain the integrity after undergoing 51 thermal shock cycles between 1100 °C and room temperature. Whereas, YSZ fiber composites burst immediately after only one thermal shock cycle under the same condition. In addition, the ZAT fiber composites also exhibit considerable mechanical and thermal insulating performance.


Introduction 
Yttria-stabilized zirconia (YSZ) ceramics are endowed with high melting point, low thermal conductivity, high chemical resistance, and considerable mechanical strength, making them promising thermal protection and energysaving materials [1][2][3][4]. YSZ fibers not only exhibit the above advantages, but also are flexible and self-supporting compared to their bulk counterparts [5,6]. These fibers can be further processed into composites, such as bricks, papers, blankets, and felts, offering additional advantages of lightweight materials and improved thermal insulation 2 Experimental details 2
Subsequently, we dissolved 5 g of PAAT precursor and 0.05 g of polyethylene oxide (PEO, M w = 1,000,000, Shanghai Aladdin Chemical Reagent Co., Ltd.) in 10 mL of anhydrous ethanol (C 2 H 5 OH, Tianjin Fuyu Fine Chemical Co., Ltd.). The mixtures were magnetically stirred for 8 h to form the transparent golden-yellow PAAT spinning solution. Then, the PAAT precursor fibers were prepared by electrospinning method with the voltage of 12 kV, the injection pump speed of 1.5 mL/h, and the distance of 20 cm at a temperature of 20-40 ℃ and relative humidity of 20%-40%. In the end, the obtained PAAT precursor fibers were heattreated at 1200-1500 ℃ for 2 h with a heating rate of 2 ℃/min in air to obtain the Al 2 TiO 5 fibers.

2 Preparation of YSZ-Al 2 TiO 5 fiber composites
The YSZ fibers (Fig. S1 in the Electronic Supplementary Material (ESM)) were prepared from polyacetylacetonatozirconium (PAZ) and Y(NO 3 ) 3 6H 2 O, and the details of the preparation procedure can be found in Ref. [35].
The as-obtained Al 2 TiO 5 and YSZ fibers were mixed with a high-speed agitator using a 2 wt% PEO aqueous solution as a low-temperature binder. The mass ratios of YSZ fibers/Al 2 TiO 5 fibers were 100/0, 99/1, 98/2, 95/5, and 92/8 (Table S1 in the ESM). The mixed fibers were injection molded under a certain pressure, sintered at 800 ℃ with a heating rate of 5 ℃/min, and then heated to 1500 ℃ for 2 h at 10 ℃/min to obtain the ZAT fiber composites.

Thermal decomposition
The molecular structure analysis of the PAAT precursor and preheated fibers was conducted with a Bruker ALPHA Fourier transform infrared (FT-IR) spectrometer in the range of 500-4000 cm −1 . Thermogravimetry and differential scanning calorimetry (TG-DSC) of the PAAT precursor fibers were performed in air with a Diamond TG-DSC Perkin Elmer thermal analyzer within the temperature range of room temperature to 1500 ℃ and the heating rate of 10 ℃/min.

Crystallization process and phase analysis
The crystalline phase compositions of Al 2 TiO 5 fibers and ZAT fiber composites were analyzed by X-ray diffraction (XRD) with a Panalytical X'pert3 powder diffractometer (40 kV, 40 mA), using Cu Kα radiation (wavelength λ = 0.15418 nm) within the 2θ range of 10°-90° and a step size of 0.02°. The thermal stability of Al 2 TiO 5 fibers was studied in Ar atmosphere by cyclic in-situ high-temperature X-ray diffraction (HTXRD) with Panaco Empyrean X-ray diffractometer, the Netherlands. The test temperature range was 1000-1500 ℃ and the heating rate and cooling rate were 5 ℃/min. The sample was scanned in the 2θ range of 10°-90°. The crystal face spacings were calculated by the Bragg equation: The lattice constants of each crystal axis were calculated according to the spacing formula for the crystal planes of the orthorhombic crystal systems, as follows:

Microscopy characterization
The microstructure of the fibers and the cross-linked state between the fibers have an important influence on the properties of composites. For microstructural characterization investigations, the Hitachi S-4800 scanning electron microscopy (SEM) equipped with an energy dispersive spectrometer (EDS) was carried out. In this process, the samples were placed on a carbon double-sided tape and coated with Au powder to make the surface electrically conductive.

Thermal expansion property
For analyzing the thermal expansion properties of the Al 2 TiO 5 fibers and the ZAT fiber composites, the thermal expansion coefficients of samples were measured with an RPZ-03P automatic thermal expansion instrument (Sinosteel Luoyang Refractory Research Institute Co., Ltd.). The sample dimension is Φ10 mm × 50 mm and the measurement condition is in the temperature range of 500-1200 ℃ and the heating rate of 5 ℃/min.

Thermal shock behavior
To investigate the thermal shock resistance behavior, the ZAT samples with size of 30 mm × 30 mm × 8 mm were quenched in water and assessed. The samples were rapidly heated from room temperature to 1100 ℃ and held for 10 min to make it heat uniformly, and then immersed in cold water immediately for rapid cooling. When the sample was cooled to room temperature, it was reheated to 1100 ℃ rapidly, and the cycles were repeated. The maximum number of thermal shock cycles at sample cracking was taken as the measurement index of thermal shock resistance.

Mechanical property
The compressive strength and the three-point bending strength of the ZAT fiber composites were measured with a DDL10 microcomputer-controlled electronic universal testing machine (Changchun Institute of Mechanical Science Co., Ltd.) with a crosshead speed of 2 mm/min. The sample dimensions are 38 mm × 38 mm × 25 mm and 100 mm × 20 mm × 20 mm for compressive strength measurement and three-point bending strength measurement, respectively.

Thermal insulation performance
The high-temperature thermal insulation performances of the ZAT samples (thickness: 8 mm) exposed to a butane blowtorch flame were characterized. The surface near the butane blowtorch flame served as the hot surface, and the surface away from the flame served as the cold surface. When the temperature of the cold surface does not change, it is considered as thermal equilibrium. Both hot surface and cold surface temperatures were intuitively observed by using an FLIR T540 infrared camera observation. The thermal conductivities of the samples with dimension of Φ12.6 mm × 1 mm were measured with the laser flash thermal conductivity instrument NETZSCH LFA 457 in the temperature range of 300-1000 ℃.  precursor and Al 2 TiO 5 fibers which were obtained by heat-treating the PAAT precursor at different temperatures. The vibration peaks at 771 and 686 cm −1 corresponded to ν(Al-O-Al) [36], while those at 659 and 492 cm −1 corresponded to ν(Ti-O-Ti) [37]. In addition, the peaks around 583 and 439 cm −1 were assigned to the hetero metal-oxygen bonds of ν(Ti-O-Al) [38]. Therefore, it could be determined that the prepared precursor was an M-acetylacetone (M = Al, Ti) complex. The peaks located at 1200-1600 cm −1 in the PAAT precursor belonged to the characteristic vibration peaks of acetylacetone [39,40]. It is obvious that with the increase of heat-treatment temperature, the intensity of absorption peaks in the 1200-1600 cm −1 region decreases a lot, indicating the decomposition of the organics.
To further analyze the thermal decomposition process of the PAAT precursor fibers in air, the TG-DSC was characterized as shown in Fig. 1(b). From the TG curve, it could be seen that evaporation of solvent and removal of the uncoordinated ligands occurred below 150 ℃, corresponding to an 8% weight loss [41]. A huge weight loss (43.5%) between 150 and 300 ℃ was caused by the decomposition of the ligands in the precursor [42]. The broad exothermic peaks in the DSC curve at 300-400 ℃ were due to the carbonization of organics and the oxidative removal of residual carbides, accompanied by a weight loss of about 17% at 300-600 ℃ in the TG curve [36,43]. When the temperature was over 600 ℃, no apparent weight loss was found. However, two exothermic peaks around 710 ℃ appeared in the DSC curve were attributed to the crystallization of rutile and corundum, and an endothermic peak at 1300 ℃ indicated the beginning of the transformation of Al 2 O 3 and TiO 2 to Al 2 TiO 5 , which was confirmed by the XRD spectra given in Fig. 1(c). In the XRD spectra, the Al 2 TiO 5 phase (PDF Card No. 26-0040) began to form at 1300 ℃. As the temperature increased, the diffraction peaks of Al 2 TiO 5 became the main crystal phase. When the temperature reached 1500 ℃, only the diffraction peaks of Al 2 TiO 5 were observed, and the peaks of Al 2 O 3 and TiO 2 completely disappeared, suggesting that the single-phase Al 2 TiO 5 fibers were successfully obtained.
In order to investigate the phase stability of the Al 2 TiO 5 fibers, the cyclic in-situ HTXRD curve of Al 2 TiO 5 fibers after heat-treatment at 1500 ℃ was shown in Fig. 1(d). During the heating cycle process in the temperature range of 1000-1500 ℃, phase change was not observed, which indicated that the Al 2 TiO 5 fibers had excellent phase stability at high temperatures. Furthermore, the unique low-thermal-expansion performance of Al 2 TiO 5 was attributed to the high thermal expansion anisotropy and the different expansion of each crystal axis [44]. The three most intense diffracted crystal planes (110), (023), and (020) were selected from the HRXRD spectrums at different temperatures in Fig. 1(d), and three crystal face spacings were calculated by Eq. (1). Because Al 2 TiO 5 is obtained as an orthorhombic crystal, the lattice constants of each crystal axis were calculated by Eq. (2), as shown in Fig. 1(e). The linear thermal expansion rates of the three axes of the Al 2 TiO 5 fibers were calculated according to ∆L/L ( Fig. 1(e)). As can be seen, the lattice constants of the b-axis and c-axis in the Al 2 TiO 5 fibers increased with increasing temperature and decreased with decreasing temperature. This is in line with the crystal thermal expansion theory, where the higher the temperature, the greater the adjacent particles' average distance, increasing the lattice constant and linear thermal expansion rate [45]. This is mainly because in the Al 2 TiO 5 crystal, Al 3+ and Ti 4+ are randomly distributed in the lattice of M metal ions, forming an [MO 6 ] (M = Al, Ti) octahedron with O 2− , as shown in the inset in Fig. 1(e) [46,47]. The commonsided octahedrons formed a double-chain structure in the b and c directions, and three coaxial octahedrons formed a single-chain structure on a-axis (a = 3.591 Å, b = 9.429 Å, c = 9.636 Å) [48][49][50].
Moreover, a fiber block was constructed to measure the thermal expansion coefficient of the Al 2 TiO 5 fibers. As shown in Fig. 1(f), the thermal expansion coefficient increased slightly with the increasing temperature. The average thermal expansion coefficient of the Al 2 TiO 5 sample was only 1.02×10 −6 K −1 from 200 to 1100 ℃, revealing that the prepared Al 2 TiO 5 fibers exhibited a low thermal expansion coefficient.
Figures 2(a)-2(c) are the macroscopic morphologies of the PAAT precursor fibers, with superior flexibility, which could unfold and recover after repeated folding. In addition, the PAAT precursor fibers had a large aspect ratio and a rather smooth surface, as suggested by the SEM images in Fig. 2(d). Figures 2(e)-2(h) display the SEM images of Al 2 TiO 5 fibers sintered at 1200, 1300, 1400, and 1500 ℃, respectively. As shown in the inset in Figs. 2(e)-2(h), the grain size gradually increased with the increase of temperature. At 1500 ℃, the fibers were formed by crystalline particles tightly connected  www.springer.com/journal/40145 and appeared the obvious sintering necks, but still maintained the one-dimensional fiber morphology. Therefore, Al 2 TiO 5 fibers offered good flexibility after heat-treatment at 1200 ℃ and became brittle after calcination at 1500 ℃, due to the obvious growth of grains in the fibers. The selected area EDS mapping results of the Al 2 TiO 5 fibers heat-treated at 1500 ℃ were displayed in Figs. 2(i)-2(l), showing that Al, Ti, and O elements of the fibers were detected. All the adopted elements were homogeneously distributed without segregation, implying that Al and Ti elements in the fibers were uniformly dispersed [51]. The Al:Ti atomic ratio was close to 2 : 1, which was consistent with the stoichiometric ratio of Al 2 TiO 5 .

2 Modification of the YSZ-Al 2 TiO 5 fiber composites
YSZ-Al 2 TiO 5 (ZAT) fiber composites were prepared by mixing the YSZ fibers and Al 2 TiO 5 fibers and injection molding followed by sintering at 1500 ℃ ( Fig. 3(a)). Figure 3(b) shows the optical images of the prepared ZAT fiber composites, and Table S1 in the ESM lists the mass, volume, density, and shrinkage rate of the ZAT fiber composites. As suggested, after heat treatment, the density and shrinkage rate of the ZAT fiber composites increased as the Al 2 TiO 5 fiber content increased. Figures 3(c)-3(l) present the SEM images of the fracture surfaces of the ZAT fiber composites, indicating that the bond between fibers was more tightly with an increasing amount of Al 2 TiO 5 fiber content, which can be attributed to the obvious sintering of the Al 2 TiO 5 fibers at high temperatures ( Fig. 2(h)). According to the research by Zu et al. [52], Ti 4+ weakened the ionicity of the Zr-O bond, allowing Zr 4+ to move easily. During the sintering process, densification enhancement was attributed to the viscous flowability of the liquid phase, strong cation diffusion, and cavity elimination ability. Additionally, the diameter of the Al 2 TiO 5 fibers was only 1-2 μm   Fig. 2(h)) and the diameter of the YSZ fibers was at least 5 μm (Fig. S1 in the ESM). The smaller the fiber diameter, the easier it sintered. Therefore, the ZAT fiber composites after heat-treatment at 1500 ℃ occurred the interfibrous sintering phenomenon.
Figures 3(m)-3(q) present the elemental mappings of ZAT-8 fiber composites, showing that the fiber composites contained O, Zr, Y, Ti, and Al elements. In the selected area, the Zr and Y elements were uniformly distributed in the coarser fibers, while Ti and Al elements were enriched in the finer fibers, indicating that the two fibers were YSZ fiber and Al 2 TiO 5 fiber. The boundary of the elemental distribution region between Zr-Y and Ti-Al was clear, demonstrating that there was no interdiffusion of multiple elements between the individual fibers, which provided strong evidence for distinguishing the Al 2 TiO 5 fibers in the YSZ fibers.
In order to further explore the phase compositions of the ZAT fiber composites, we analyzed the XRD patterns shown in Fig. 4(a), showing that the ZAT fiber composites were composed of a ZrO 2 phase (PDF Card No. 89-9069) and an Al 2 TiO 5 phase (PDF Card No. 26-0040). As the Al 2 TiO 5 fiber content increased, the diffraction peak intensity of the Al 2 TiO 5 phase became higher, which could be explained by the absence of chemical bonding between the YSZ fibers and the Al 2 TiO 5 fibers.
The low thermal expansion coefficient of the Al 2 TiO 5 fibers could be partially offset the thermal expansion coefficient of ZAT fiber composites. The measurement results showed that the thermal expansion coefficients of ZAT fiber composites were found to be negatively correlated with Al 2 TiO 5 fiber content (Fig. 4(b)). Among them, the average thermal expansion coefficient of the ZAT-0 fiber composites exceeded 10×10 −6 K −1 at 500-1200 ℃. In comparison, when the Al 2 TiO 5 fiber content was 8 wt%, the ZAT fiber composites had a minimum thermal expansion coefficient of 7.74× 10 −6 K −1 , which was about 26% lower than that of the ZAT-0 fiber composites (Fig. 4(d)). In addition, the linear thermal expansion rate of ZAT samples showed a downward trend with an increase of Al 2 TiO 5 fiber content as shown in Fig. 4(c). The average linear www.springer.com/journal/40145 expansion rate of the ZAT-0 fiber composites was measured to be 0.83%, while that of the ZAT-8 fiber composites was reduced to 0.60% (Fig. 4(d)). Therefore, the above results confirmed that the Al 2 TiO 5 fibers could effectively reduce the thermal expansion coefficient of the YSZ fiber composites, resulting in a superior low thermal expansion property of the ZAT fiber composites.
To further verify that the reduction in thermal expansion coefficient was beneficial to improve the thermal shock resistance of the ZAT fiber composites, the thermal shock behaviors of the ZAT fiber composites under repeating thermal cycling were studied using the water-quenching method. The ZAT fiber composites with size of 30 mm × 30 mm × 8 mm were rapidly heat-treated to 1100 ℃ and then immersed in cold water immediately for rapid cooling (Fig. 5(a)). The morphological characteristics of the ZAT samples after multiple thermal shock cycles were summarized in Table S2 in the ESM. As shown in Fig. 5(b), the ZAT-0 fiber composites burst and broke instantaneously after only one thermal shock cycle. By contrast, the thermal shock resistance of the ZAT fiber composites improved significantly (Figs. 5(c)-5(f)), and the higher Al 2 TiO 5 fiber content led to better thermal shock resistance of the ZAT fiber composites. In particular, the ZAT-8 sample only cracked slightly after undergoing 51 thermal shock cycles, exhibiting outstanding thermal shock resistance (Fig. 5(f)). It can be concluded that the addition of Al 2 TiO 5 fibers could effectively enhance the thermal shock resistance of the YSZ fiber composites. The reason is that according to the thermal shock fracture initiation and crack propagation theory proposed by Hasselman, the stability factor of the thermal stress increases as the thermal expansion coefficient decreases, making it more difficult for the crack to propagate, resulting in an improvement in the thermal shock resistance of the material [53,54].
The mutual connections between the YSZ fibers and Al 2 TiO 5 fibers were expected to increase the mechanical properties of the ZAT fiber composites. Thus, the mechanical properties of the ZAT fiber composites were investigated (Fig. 6). As expected, the compressive strengths of the prepared ZAT fiber composites improved with the increase of Al 2 TiO 5 fiber content can be clearly observed in Figs. 6(a) and 6(b). Among them, the maximum compressive strength of the ZAT-8 fiber composites was measured to be 20.35 MPa, which was about 18 times larger than that of the ZAT-0 fiber composites (1.14 MPa). The crack propagation paths during fracture of the ZAT fiber composites determined a ductile fracture mode, according to the compressive stress-strain curves of the ZAT fiber composites in Fig.  6(c). Furthermore, Young's modulus of the ZAT fiber composites increased with increasing Al 2 TiO 5 fiber content ( Fig. 6(b)). Of note, Young's modulus of the ZAT-8 fiber composites reached 13.92 MPa, which was much higher than that of the ZAT-0 fiber composites. To comprehensively analyze the mechanical properties of the ZAT composites, the three-point bending tests were also carried out as shown in Figs. 6(d) and 6(e). Similarly, the three-point bending strength of the ZAT-8 fiber composites was 56.52% higher than that of ZAT-0 fiber composites. As suggested, the bending stress increased almost linearly with the displacement under the bending loads as shown in Fig. 6(f). In addition, the  compressive strength and three-point bending strength of the ZAT fiber composites were higher than those of ZAT-0 fiber composites at the same density displayed in Figs. 6(a) and 6(c). Investigations revealed that the Al 2 TiO 5 fibers could notably modify the mechanical properties of the YSZ fiber composites, which is due to the sintering and crosslinking between the intersection of the YSZ fibers and Al 2 TiO 5 fibers at high temperatures.
Because ZAT fiber composites have been primarily applied to thermal insulation industries, the thermal insulation performance at high temperature was measured by subjecting the ZAT samples to a butane blowtorch flame. A detailed schematic of the device is depicted in Fig. 7(a), and the sample size is 30 mm × 30 mm × 8 mm (Fig. 7(b)). Figure 7(c) shows that the maximum temperature of the hot surface could reach 1300 ℃, while the cold surface temperature of the ZAT-0 fiber composites was stable at 552 ℃ after 300 s (Fig. 7(d)). In Figs. 7(d)-7(g), when the Al 2 TiO 5 fiber content was less than 5 wt%, the cold surface temperature of the ZAT fiber composites exhibited an obvious decreasing trend with increasing Al 2 TiO 5 fiber content. In particular, ZAT-5 fiber composites stabilized at 464 ℃ as shown in Fig. 7(g), indicating excellent thermal insulation performance. Nevertheless, the comparisons showed that when the Al 2 TiO 5 fiber content increased to 8 wt%, the cold surface temperature increased to 480 ℃ given in Fig. 7(h). The cold surface temperatures of the ZAT fiber composites first decreased and then increased with increasing Al 2 TiO 5 fiber content, as shown in Fig. 7(i). Furthermore, Fig. 7(j) displays the thermal conductivity of the ZAT fiber composites as a function of temperature. Similarly, with the increase of the Al 2 TiO 5 fiber content, the thermal conductivity of the ZAT fiber composites first decreased and then increased, and the ZAT-5 fiber composites had the lowest thermal conductivity of 0.555 Wm −1 ·K −1 at 1000 ℃. These results indicated that at similar densities, the thermal insulation performance of the ZAT fiber composites improved with increasing Al 2 TiO 5 fiber content. However, when the density increased to a certain value, the thermal conductivity increased as the density increased [55]. Thus, the thermal insulation performance of the ZAT-8 fiber composites was slightly decreased due to the higher density. In addition, the thermal conductivities of the YSZ and Al 2 TiO 5 fibers with similar densities were tested (Fig. 7(k)). At the same temperature, the thermal conductivity of the Al 2 TiO 5 fibers was noticeably lower than that of the YSZ fibers, which demonstrated the superior thermal insulation of the Al 2 TiO 5 fibers.
As a result, the addition of Al 2 TiO 5 fibers enhanced the thermal shock resistance and mechanical properties of the YSZ fiber composites, as well as ensured thermal insulation performance. Appropriate Al 2 TiO 5 fiber content (≤ 5 wt%) reduced the thermal conductivity of the ZAT fiber composites, while excessive Al 2 TiO 5 fiber content (8 wt%) greatly increased the density of the fiber composites, resulting in a decline in thermal insulation performance. With higher Al 2 TiO 5 fiber content, such as 12 wt% (denoted as ZAT-12) and 16 wt% (denoted as ZAT-16), the thermal shock resistance continued to increase slightly (Fig. S3 in the ESM) due to the reduced thermal expansion coefficient of the ZAT fiber composites (Fig. S2 in the ESM). Both these materials had higher densities than the composites with low Al 2 TiO 5 fiber content (Table S1 in the ESM), resulting in increased mechanical strength (Fig. S4 in the ESM). As expected, the thermal insulation performances of ZAT-12 and ZAT-16 fiber composites were inferior to that of ZAT-8 fiber composites with increasing densities (Fig. S5 in the ESM). Comprehensive consideration of the mechanical and thermal properties, the content of Al 2 TiO 5 fibers as 8 wt% was the optimum proportion.

Conclusions
In this work, YSZ-Al 2 TiO 5 fiber composites were fabricated for the purpose of enhancing the thermal shock resistance of YSZ fiber composites. The Al 2 TiO 5 fibers were incorporated into YSZ fiber composites by directly mixing the two fibers followed by compression molding and heat treatment. With the addition of 8 wt% of Al 2 TiO 5 fibers, the thermal expansion coefficient of the sample was as low as 7.74×10 −6 K −1 , which was 26% lower than that of the YSZ fiber composites. The modified YSZ fiber composites underwent 51 thermal shock cycles between 1100 ℃ and room temperature, compared to only one thermal shock for the pure YSZ fiber composites. The mechanical properties of the composites increased remarkably with increasing Al 2 TiO 5 fiber content, and the compressive strength was as high as 20.35 MPa. The modified fiber composites were capable of isolating a hot surface temperature of 1300 ℃ and stabilizing the cold surface temperature at 464 ℃ with a thickness of 8 mm. Our findings may provide new opportunities for expanding the application fields of YSZ fiber composites as well as provide novel guidance for the rational structural design of fiber composites.