June 2016 Vol.59 No.6: 913–919 doi: 10.1007/s11431-015-5978-x
Improving the thermal shock resistance of ceramics by crack arrest blocks WANG YanWei1,2, XIA Biao1,2, SU HongHong1,3, CHEN Hang1,2 & FENG Xue1,2* 1
Applied Mechanics Laboratory, Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China; 2 Center for Mechanics and Materials, Tsinghua University, Beijing 100084, China; 3 Department of Engineering Mechanics, Chongqing University, Chongqing 400044, China Received July 6, 2015; accepted October 20, 2015; published online February 19, 2016
Ceramics used in the high temperature environment are inevitably subjected to sudden temperature change, which may lead to catastrophic thermal shock failure due to the intrinsic brittleness of ceramics. In this paper, an experimental platform is designed to realize the in-situ observation during the thermal shock experiments. Experimental results show that all the cracks initiate from one of the edge midpoints and propagate to another one for square specimens. Such experimental observation is consistent with the maximum tensile stress zone with the maximum temperature gradient given by the finite element method (FEM). The different crack modes resulting from different heating rates after thermal shock experiments are observed and analyzed. Comparison between different clamping methods is conducted to study the effects of boundary conditions on the thermal shock experiments. Furthermore, in order to improve the thermal shock performance of alumina ceramics, crack arrest blocks are added near the edge midpoint. The thickness, shape and arrangement of the blocks are systematically investigated to understand the mechanism of improvement of thermal shock resistance. thermal shock, ultra-high temperature ceramics, crack arrest blocks, in-situ dynamic observation Citation:
Wang Y W, Xia B, Su H H, et al. Improving the thermal shock resistance of ceramics by crack arrest blocks. Sci China Tech Sci, 2016, 59: 913−919, doi: 10.1007/s11431-015-5978-x
1 Introduction In recent years, researchers and engineers have made increasing efforts to improve the performance of materials of aero-engine components [1,2] due to aircrafts with high Mach number. Due to the high melting point, better corrosion resistance and wear resistance, ceramics are extensively used in aerospace industry. However, the ceramics are prone to experience thermal shock failure due to their intrinsic brittleness . Moreover, for the serving environments with high temperature, the outer ceramic layers on the components suffer from both thermal shock and ablation
[4–6]. Therefore, the thermal shock resistance of ceramics becomes one of the important criteria in the selection and design of aerospace materials . According to the temperature difference, thermal shock can be divided into two categories: hot shock and cold shock. For hot shock, the specimen is directly heated to a target temperature [8–13] with a given heating rate. For cold shock, specimens are firstly heated to a target temperature, maintained for a certain time, and quickly put into the cooling medium such as water [14,15], oil , liquid metal , and gas [18,19]. During the thermal shock experiments, cracks are generated in the specimens when the thermal stress exceeds their strength and the maximum temperature difference at the occurrence of cracks is used to characterize
the thermal shock resistance of the specimens. Cold shock, especially water quenching method, is generally adopted in traditional thermal shock test experiments due to the fact that hot shock test is limited by temperature increasing rate. Hasselman  studied the crack propagation behavior of cylindrical polycrystalline alumina rods with water quenching methods. It was found that the strength decreased dramatically when crack initiated, and remained constant until its further propagation or formation of new ones. Bahr et al. [21,22] and Shao et al. [23,24] studied the single and multiple crack patterns in ceramic plates after water quenching, and analyzed the distribution of interfacial cracks. However, the experimental conditions of traditional water quenching method are quite different from the real environment in aerospace, where materials are often subjected to the radiation or thermal shock during the rapid and non-uniform heating process with flame. Moreover, it is difficult to achieve in-situ dynamic observation for rapid water quenching due to the complex physical and chemical reactions during the process. The most common strategy to improve the thermal shock resistance is to add reinforcements into the ceramics, such as nickel, molybdenum, silicon nitride, graphite flake and silicon carbide [25–28], to optimize the lattice structure of the material. By the reduction of thermal resistance between ceramic grain boundaries, Li et al.  showed that the thermal shock resistance of alumina ceramic can be increased from 300 to 650°C. However, the improvement of the thermal shock resistance always means the sacrifice of other material properties of the materials. The thermal shock resistance of the ceramics is not only related to their thermal properties, but also closely related to the interface structure of the ceramic materials [29–34]. Based on the microstructure of the wing membrane of dragonfly, Song et al.  proposed a new method to pattern the surface of ceramics to a biomimetic nano-finned surface. During quenching, a thin air layer is created to envelop the surface, which enhances the thermal resistance of the structure. However, it is difficult to fabricate such micro-structures onto the surface of such brittle and chemically inert ceramic, and theoretical method is also scarcely proposed to guide the design and optimization of their shape and alignment. In this paper, an experimental platform is designed to realize the in-situ crack observation during the thermal shock experiments, in which the oxy-propane flame is used to heat the ceramics with the heating rate up to 150°C/s. Thermal shock experiments are performed for square ceramic specimens with different heating speeds, which lead to different crack modes. FEM is employed to reveal the stress and temperature distribution in the specimen in order to explain the laws for cracks initiation. Square and semicircular crack arrest blocks with different sizes are added to the edge midpoints of each specimen to improve the thermal shock resistance, which is demonstrated by experiments and FEM.
June (2016) Vol.59 No.6
2 Materials and methods Square ceramic specimens (alumina 96%) are prepared with tape casting process, with the size of 35 mm×35 mm×1 mm, the surface roughness of 0.2–0.75 µm, and the average grain size of 3–5 µm. The properties of ceramics at room temperature are shown in Table 1. Figure 1 shows the diagram of experimental setup for dynamic thermal shock, where the specimen is clamped at a corner by a V-shaped holder. The center of the specimen is heated with the oxygen-propane flame via automatic control program with the surface of the specimen perpendicular to the flame direction from the nozzle. The fixed output pressures and flow rates of oxygen and propane are 0.5 MPa, 100 L/h and 0.09 MPa, 60 L/h respectively. The generation and propagation of the thermal shock cracks are observed by high-speed camera (Phantom v1610). The temperature increase at the center of the rear side is measured by infrared thermometer pyrometer (Raytek MI3). Heating rate applied to the center of the specimen surface is adjusted by changing the distance between the nozzle and the specimen. Experiments are also performed to eliminate the effects of boundary conditions. Three different clamping methods are tested: 1) Clamping at one corner, 2) clamping at one edge and 3) constraining by a network. To depress the propagation of the cracks, crack arrest blocks are fabricated to the crack propagation path. We propose to add crack arrest blocks surrounding the zones Table 1
Physical properties of alumina ceramics
Elastic modulus (GPa)
Tensile strength (MPa)
Thermal expansion coefficient (K−1) 6.5–8.0×10−6
Figure 1 (Color online) Photo of the in-situ dynamic thermal shock experiment platform. (1) The oxygen-propane flame heating system; (2) specimen holder; (3) high-speed camera; (4) LED cold light source for high-speed camera; (5) infrared thermometer pyrometer.
Wang Y W, et al.
Sci China Tech Sci
with maximum tensile stress (i.e. midpoints of each edge), which will be discussed in the next section, to improve the thermal shock resistance of ceramics. As shown in Figure 2, four square or semicircular crack arrest blocks (the material are the same as the specimen) are adhered to the edge midpoints on the heating surface (loading surface) by AK04-4 high-temperature inorganic adhesive glue. Two kinds of square blocks with the same in-plane size 10 mm × 10 mm (Figure 2), but different thickness h are prepared with 1 mm (i.e. thin blocks) and 3 mm (i.e. thick blocks) thickness respectively.
3 Results and discussion After the thermal shock experiments, three different failure modes of the square specimens are observed corresponding to different heating rates, i.e. single crack, herringbone crack and the combination of single and herringbone cracks, as shown in Figure 3. The number of cracks increases with the increasing of heating rate to release the thermal shock energy. However, the cracks of all the above modes initiate from the midpoint of each specimen edge and propagate through two midpoints. The reason is that, along the edge, the midpoint is the nearest point to the heating source. It indicates that the direction along the edge midpoint and the center is consistent with that of the maximum temperature gradient direction in the specimen. The steady-state thermal-mechanical coupling analysis of the thermal shock is also performed by ABAQUS  to verify the above qualitative explanation. The size, properties of the materials and boundary conditions at the clamped part are set to be consistent with experiments. The specimen is meshed by C3D8T elements. The predefined temperature field is set as room temperature for all nodes. The air convection is simulated by defining an interactive behavior to the surface of the specimen. The surface heat flux (9×105 W/m²) is applied to the central zone with the radius of 2 mm measured by experimental observation. Figure 4(a) shows the transient temperature increase in the specimen with the heating power 11.3 W for 10 s, where the contour presents as concentric circles. The maximum temperature gradient is indeed located in the direc-
Figure 2 (Color online) Image and schematic diagram of specimens with crack arrest blocks: (a) Image captured by high-speed camera; (b) schematic diagram.
June (2016) Vol.59 No.6
tions from the center of the specimen to four edge midpoints. Figure 4(b) shows the maximum principle stress σ1 distribution in the same specimen, where the maximum value of the specimen is tensile stress and located in the edge midpoints. Since ceramics are more sensitive to tensile stress, it is reasonable to adopt the maximum value of σ1 as the criterion of crack initiation. Therefore, all the cracks initiate from one of the edge midpoints, and propagate to another one to minimize the energy of the system. As shown in Figure 5, for the same heating rate, the specimens are (a) clamped at one corner and one edge and (b) clamped at one corner and constrained by a network presenting the same failure mode respectively. This result indicates that the boundary conditions have negligible effects on the thermal shock experiments. Compared with the specimens without crack arrest block, the specimens with square crack arrest blocks present completely different thermal shock failure modes. For thin blocks, some cracks initiate from the edge midpoint, while others initiate from the edge of the blocks, as shown in Figure 6(a). This is because the maximum value of σ1 in the specimen is also located on the edge midpoint before the cracks initiation, but the distribution of σ1 is more uniform along the edge, as shown in Figure 7(a). On the other hand, for thick blocks, all the cracks no longer initiate from the edge midpoints (Figure 6(b)) because the maximum value of σ1 in the specimen is relocated at the edge of the blocks,
Figure 3 (Color online) Thermal shock failure modes of the square specimens: (a) Single crack (428°C), (b) herringbone crack (456°C) and (c) the combination of single and herringbone cracks (476°C). The temperature in the brackets is given by the infrared thermometer pyrometer when the cracks initiate.
Figure 4 (Color online) FEM results of the square specimen under thermal shock: (a) Temperature contour; (b) maximum principal stress contour.
Wang Y W, et al.
Sci China Tech Sci
Figure 5 (Color online) Comparison of specimen under thermal shock with different boundary conditions: (a) Clamped at one corner and one edge, and (b) clamped at one corner and constrained by a network.
June (2016) Vol.59 No.6
load intensity, the critical temperature corresponding to the failure of the specimen increases with crack arrest blocks thickness h, where h=0 represents specimen without crack arrest blocks. It is demonstrated again that thicker blocks could improve higher thermal shock resistance for alumina ceramics. In order to investigate the effects of block shape, different semicircular crack arrest blocks are studied by FEM. The radius of the semicircular blocks R=8.0 mm is determined by the same in-plane area as the square blocks. For the same thickness (h=3 mm), Figure 10 shows the comparison of the maximum σ1 in the specimens with square and semicircular blocks, which indicates that the square blocks have better performance to improve the thermal shock resistance for materials with relatively low tensile strength, while the semicircular blocks are better for materials with relatively high tensile strength. Although the crack arrest blocks above are added to the same side with the heating flame, altering the heating flame side (i.e. heating at the side without blocks) has negligible effects on the thermal shock
Figure 6 (Color online) Thermal shock failure modes on the observation surface of specimens with square crack arrest blocks: (a) Thin blocks and (b) thick blocks.
as shown in Figure 7(b). Comparing Figure 6(a) and (b), it has been also found that, under the same thermal load intensity, specimens with thin blocks fractured into four parts, while those with thick blocks only broke into three parts. Therefore, thicker blocks can absorb more thermal shock energy to prevent the crack propagation in the specimen. Figure 8 shows the maximum value of σ1 in the specimen as the thermal shock time. The stress evolution of the specimen with thin blocks presents the same tendency as that without blocks due to the same dangerous point but different from that with thick blocks. The maximum value of the specimen with thick blocks is larger than that with thin blocks at the beginning of thermal shock, while it turns to be less than that with thin blocks after a critical thermal shock time. In other words, with the increasing of blocks thickness, the stress distribution becomes more uniform, which delays the crack initiation and propagation providing positive effects on the thermal shock resistance for materials with high tensile strength. Meanwhile, as shown in Figure 9, for a given thermal
Figure 7 (Color online) Maximum principal stress distribution of the specimens added with various square crack arrest blocks: (a) Thin blocks and (b) thick blocks.
Figure 8 (Color online) The maximum value of σ1 in the specimens with square crack arrest blocks of different thickness.
Wang Y W, et al.
Sci China Tech Sci
Figure 9 The failure temperatures of square specimens under thermal shock with square crack arrest blocks of different thickness.
Figure 10 (Color online) The comparison of the maximum σ1 in the specimens with square and semicircular blocks.
performance of the specimen, as demonstrated by the comparison of maximum σ1 in the specimens before and after altering the heating flame side in Figure 11. However, if the blocks are divided into two parts and arranged symmetrically on both sides of the specimens, the improvement of the thermal shock resistant will be decreased, as demonstrated in Figure 12, where the maximum σ1 in the specimen with semicircular blocks on one side (h=3 mm) is smaller than that with blocks symmetrically on both sides (h=1.5 mm). This is because the blocks symmetrically arranged on the both sides have stronger constraints for the part of specimen between them, where the stress level is always low along the thickness direction, so that the deformation energy is isolated in the area without blocks, as shown in Figure 13(b). On the contrary, taking advantage of unsymmetrical arrangement, the stress distribution on the other side is more uniform due to bending deformation, as shown in Figure 13(a).
June (2016) Vol.59 No.6
Figure 11 (Color online) The comparison of the maximum σ1 in the specimens with the blocks (h=1.5 mm) on the same and contrary sides as the heating flame.
Figure 12 (Color online) The comparison of the maximum σ1 in the specimens with semicircular blocks arranged symmetrically on both sides and unsymmetrically on one side with the same total thickness 3 mm.
Figure 13 (Color online) The maximum principle stress contour in the specimens with semicircular blocks arranged (a) unsymmetrically on one side and (b) symmetrically on both sides and with the same total thickness 3 mm.
Wang Y W, et al.
Sci China Tech Sci
The circular specimens are also analyzed by experiments. During the thermal shock experiments, the cracks randomly present along the diameter of the specimen due to its strong rotational symmetry. In this circumstance, the crack arrest blocks cannot obviously improve the thermal shock performance.
In this paper, based on the experimental platform with the in-situ observation, the thermal shock failure crack modes of square alumina ceramic specimens under the central flame heating have been systematically studied. The experiment results show that, for the square specimens, all the cracks initiate at the midpoint of the edges and propagate to another one, which is consistent with the maximum tensile stress zone with the maximum temperature gradient given by FEM. Through comparison between different clamping methods, it is demonstrated that boundary conditions have negligible effects on the thermal shock experiments. The experiments and FEM results show that crack arrest blocks can significantly improve the thermal shock resistance by inducing redistribution of stress to relocate the crack initiation points, which provide a practical method for crack arresting. And thicker blocks could achieve higher thermal shock resistance for alumina ceramics. FEM results also indicate that the square blocks have better performance to improve the thermal shock resistance for materials with relatively low tensile strength and the side which be altered to add the crack arrest blocks has negligible effects on the thermal shock performance. Furthermore, for the same total thickness, the blocks arranged unsymmetrically on one side is better than symmetrically on both sides to improve the thermal shock resistance. In summery, it is demonstrated, by both the experiments and qualitative analysis, that all the cracks initiate at the maximum tensile thermal stress zone with the maximum temperature gradient. Both FEM and qualitative analysis can be employed to guide the detailed design. For square specimen, the crack arrest blocks should be located at the midpoint of the edges originally with the maximum tensile stress. On the other hand, the crack arrest blocks should be ring shape around the outer edge of the specimen.
This work was supported by the National Basic Research Program of China (“973” Project) (Grant No. 2015CB351900), the National Natural Science Foundation of China (Grant Nos. 11222220, 11320101001, 11372155 & 11227801), and the Tsinghua University Initiative Scientific Research Program. 1
Ma Y J, Yao X F, Hao W F, et al. Oxidation mechanism of ZrB2/SiC ceramics based on phase-field model. Compos Sci Technol, 2012, 72: 1196–1202 Ma Y J, Liu X Y, Yao X F, et al. Buckle-driven delamination of thermal barrier coatings on a polynomial curved substrate. Modell
June (2016) Vol.59 No.6
Simul Mater Sci Eng, 2013, 21: 055026 Kingery W D. Factors affecting thermal stress resistance of ceramic materials. J Am Ceram Soc, 1955, 38: 3–15 Wu D F, Zhou A F, Zheng L M, et al. Study on the thermal protection performance of superalloy honeycomb panels in high-speed thermal shock environments. Theor Appl Mech Lett, 2014, 4: 021004 Chen X M, Liu Z W, Yang Y. Residual stress evolution regularity in thermal barrier coatings under thermal shock loading. Theor Appl Mech Lett, 2014, 4: 021009 Lyukshin B A, Lyukshin P A, Bochkareva S A, et al. Stress-strain state and loss of stability of anisotropic thermal coating under thermal shock. AIP Conf Proc, 2014, 1623: 387–390 Hasselman D P H. Theory of thermal shock resistance of semitransparent ceramics under radiation heating. J Am Ceram Soc, 1966, 49: 103–1048 El-Adawi M K, Elshehawey E F. Heating a slab induced by a time-dependent laser irradiance-an exact solution. J Appl Phys, 1986, 60: 2250–2255 Fang X F, Jia J M, Feng X , et al. Three-point bending test at extremely high temperature enhanced by real-time observation and measurement. Rev Sci Instrum, 2014, 85: 035104 Levine S R, Opila E J, Halbig M C, et al. Evaluation of ultra-high temperature ceramics for aeropropulsion use. J Eur Ceram Soc, 2002, 22: 2757–2767 Wakayama S, Kobayashi S, Wada T. AE characterization of thermal shock crack growth behavior in alumina ceramics by Disc-on-Rod test. J Acoustic Emission, 2005, 23: 150–155 Fang X F, Yu H L, Zhang G B, et al. In situ observation and measurement of composites subjected to extremely high temperature. Rev Sci Instrum, 2014, 85: 035104 Wang Y W, Yu H L, Tang H X, et al. Crack arrest in brittle ceramics subjected to thermal shock and ablation. Chin Phys Lett, 2014, 31: 094601 Komeya K, Haruna Y, Meguro T, et al. Oxidation behaviour of the sintered Si3N4-Y2O3-Al2O3 system. J Mater Sci, 1992, 27: 5727–5734 Yoshimoto T, Ishihara S, Goshima T, et al. An improved method for the determination of the maximum thermal stress induced during a quench test. Scripta Mater, 1999, 41: 553–559 Aratani S, Sato K. Experimental study of heat transfer coefficient with temperature and kinematic viscosity on mineral oil and molten salt. J Ceram Soc Jpn Int Ed, 1995, 103: 373–377 Oguma M, Fairbanks C J, Hasselman D P H. Thermal stress fracture of brittle ceramics by conductive heat transfer in a liquid metal quenching medium. J Am Ceram Soc, 1986, 69: C-87–C-88 Faber K T, Huang M D, Evans A G. Quantitative studies of thermal shock in ceramics based on a novel test technique. J Am Ceram Soc, 1981, 64: 296–301 Johnson-Walls D, Drory M D, Evans A G, et al. Evaluation of reliability of brittle components by thermal stress testing. J Am Ceram Soc, 1985, 68: 363–367 Hasselman D P H. Strength behavior of polycrystalline alumina subjected to thermal shock. J Am Ceram Soc, 1970, 53: 490–495 Bahr H A, Weiss H J. Heuristic approach to thermal shock damage due to single and multiple crack growth. Theor Appl Fract Mec, 1986, 6: 57–62 Bahr H A, Fischer G, Weiss H J. Thermal-shock crack patterns explained by single and multiple crack propagation. J Mater Sci, 1986, 21: 2716–2720 Shao Y F, Xu X H, Meng S H, et al. Crack patterns in ceramic plates after quenching. J Am Ceram Soc, 2010, 93: 3006–3008 Shao Y F, Du R Q, Wu X F, et al. Effetc of porosity on the crack pattern and residual strength of ceramics after quenching. J Mater Sci, 2013, 48: 6431–6436 Meléndez-Martınez J J, Domınguez-Rodrıguez A, Monteverde F, et al. Characterisation and high temperature mechanical properties of zirconium boride-based materials. J Eur Ceram Soc, 2002, 22: 2543– 2549
Wang Y W, et al.
Sci China Tech Sci
Sbaizero O, Pezzotti G. Influence of molybdenum particles on thermal shock resistance of alumina matrix ceramics. Mater Sci Eng A, 2003, 343: 273–281 Monteverde F, Bellosi A. Effect of the addition of silicon nitride on sintering behaviour and microstructure of zirconium diboride. Scripta Mater, 2002, 46: 223–228 Zhang X H, Wang Z, Hu P, et al. Mechanical properties and thermal shock resistance of ZrB2-SiC ceramic toughened with graphite flake and SiC whiskers. Scripta Mater, 2009, 61, 809–812 Li W G, Cheng T B, Zhang R B, et al. Properties and appropriate conditions of stress reduction factor and thermal shock resistance parameters for ceramics. Appl Math Mech, 2012, 33: 1351–1360 Sherman D, Schlumm D. Thickness effect in thermal shock of alumina ceramics. Scripta Mater, 2000, 42: 819–825 Becher P F. Thermal shock resistance of ceramics size and geometry
June (2016) Vol.59 No.6
effects. Am Ceram Soc Bull, 1980, 59: 542 Li D J, Li W G, Li D Y, et al. Theoretical research on thermal shock resistance of ultra-high temperature ceramics focusing on the adjustment of stress reduction ractor. Mater, 2013, 6: 551–564 Jin Z H, Feng Y Z. Thermal shock resistance behavior of a functionally graded ceramic: Effects of finite cooling rate. Theor Appl Mech Lett, 2014, 4: 041008 Li B C, Mandelis A, Kish Z Z. Photothermal investigation of the thermal shock behavior of alumina ceramics for engine components. J Appl Phys, 2004, 95: 1042–1049 Song F, Meng S H, Xu X H, et al. Enhanced thermal shock resistance of ceramics through biomimetically inspired nanofins. Phys Rev Lett, A, 2010, 104: 125502 ABAQUS. Analysis user’s manual v.6.9. Pawtucket, RI: Dassault Systèmes. 2009