S. Bueno et al. /Journal of the European Ceramic Society 25(2005)847-850 when static Youngs modulus is considered, as it is much alumina layers in the laminates A10A40 and AA10, re- lower for the composite and significantly different from dy pectively. Since residual stresses reduce the strength namic Young's modulus(Table 2). The former parameter is thin (300 um) intermediate layers and relatively thick tions imposed during the tests as compared to those imposed minimise them. With such structure, negligible residual in dynamic measurements. Accordingly, the fact that the stresses would be developed in the system AAlO in which static Youngs modulus for AlO(A+T)is significantly lower high strength values are expected than the dynamic modulus would indicate that deformation For the preparation of laminates, suspensions with imposed during loading originates incipient microcracking viscosity and high solids content result in a better in the material, undetectable by a changing of the curvature trol of wall thickness formation. Casting rates of 0.53 and of the nominal stress-apparent strain curve(Fig. 3a) 1.31 mm-min(corresponding to the slopes of the kinet In the materials A30(A+T)and A40(A+T) the lower ics curves in Fig. 5)are found for alumina and A40(A+T), strength values (Table 2), large strains at fracture and respectively, which allow a good control of the casting non-linear behaviour are due to the presence of microcracks, process. The layered green bodies were fabricated by se- which originate highly tortuous fracture(Fig. 4b and c) quential slip casting. The thickness of each layer in the process. These microcracks lead to low values of the dy- taking into account the sintering shrinkage experienced by namic and static Youngs modulus and thermal expansion the monoliths with the same compositions as those of the (Table 2) layers. Table 3 summarises casting times and the compari- In order to select the most adequate material to constitute son between the obtained thickness of the layers, measured the internal faw tolerant layers in the laminates, the brittle- directly in the SEM, and the calculated ones. There are sig- ness parameter proposed by Gogotsi can be used to quan- nificant differences( 25-35%)between the calculated and tify the apparent ductility of the two composites containing the obtained widths for the layers made from the A40(A+T) the largest aluminium titanate amounts. This parameter is slips, whereas those corresponding to the AlO(A+t) slips defined by the ratio of the specific elastic energy which has are inside the uncertainty range associated with the mea been accumulated in the sample at fracture, calculated from surement method and the variability between samples. This the apparent Youngs modulus and the elastic deformation fact is explained in terms of slip quality, as the one made at fracture, to the whole specific energy expended during the of A40(A+T) mixture presented the poorest rheological test, determined as the whole area under the stress-strain behaviour, thus making the control of casting rate difficult curve( Fig. 3b). As shown in Table 2, the brittleness param- For the alumina layers, very high casting times (1000s eter is the lowest, indicating larger apparent ductility, for the for the central layer, Table 3)are necessary and therefore composite containing 40 vol. of aluminium titanate. This a possible change in the slip properties could modify the apparent ductility leads to increased energy absorption dur casting kinetics and the obtained thickness ing fracture in the expense of strength, as demonstrated by A comparison between the stress-strain behaviour of the the fact that only one-fourth of the strength of the AlO(A+t) sintered laminates and that of monolith samples of the same monolith is achieved by these samples size is established in Fig. 7. The stress-strain behaviour of the laminate A10A40 was linear up to fracture, such as that 42. laminates of the monoliths that constituted its external and central lay ers, made of 10(A+T). This fact indicates that the exter From the above-mentioned results, two laminates with nal and central uncracked layers dominate this deformation five layers were designed. The system Al0A40 was se- range. The monolith AAlO behaves also linearly, as both lected to combine low strength and energy absorbing in- constituents. Nevertheless, the slope of this linear portion termediate layers of A40(A+T)composition with sufficient was lower for the layered materials than for the large mono- strength provided by external layers of A10(A-+T)composi- lithic samples with the same composition as those of the tion, which was preferred over monophase alumina because corresponding external layers, due to the effect of the low of chemical compatibility between the layers. In the system Youngs modulus internal layers AAlO, external and central layers of monophase alumina The stress drop from the failure point of the large mono- with high strength were combined with intermediate lay- lith samples occurred steeply, and apparent deformation at ers of A10(A+T)composition, in which microcracks might the maximum load was coincident with that for zero load be developed at most during loading, as discussed above. after fracture(Fig. 7). In the laminates, a step-like way was For symmetry, both laminates were designed with internal followed and apparent strain for zero load was larger than central layers of the same composition as that of the corre- that corresponding to failure load. This fracture behaviour sponding external layer is related to the changes in the crack path at the interl Considering Youngs modulus and thermal expansion co- ers, as suggested by the fracture surfaces(Fig. 6). The ef- efficients values, given in Table 2, tensile residual stresses fect is more significant for A10A40(Fig. 7b)in which, in are expected in the external and central AlO(A+T) and addition to the crack deflection due to differences between854 S. Bueno et al. / Journal of the European Ceramic Society 25 (2005) 847–856 when static Young’s modulus is considered, as it is much lower for the composite and significantly different from dy￾namic Young’s modulus (Table 2). The former parameter is more sensible to microcracks because of the larger deforma￾tions imposed during the tests as compared to those imposed in dynamic measurements. Accordingly, the fact that the static Young’s modulus for A10(A+T) is significantly lower than the dynamic modulus would indicate that deformation imposed during loading originates incipient microcracking in the material, undetectable by a changing of the curvature of the nominal stress–apparent strain curve (Fig. 3a). In the materials A30(A+T) and A40(A+T) the lower strength values (Table 2), large strains at fracture and non-linear behaviour are due to the presence of microcracks, which originate highly tortuous fracture (Fig. 4b and c) revealing additional energy absorption during the fracture process. These microcracks lead to low values of the dy￾namic and static Young’s modulus and thermal expansion (Table 2). In order to select the most adequate material to constitute the internal flaw tolerant layers in the laminates, the brittle￾ness parameter proposed by Gogotsi16 can be used to quan￾tify the apparent ductility of the two composites containing the largest aluminium titanate amounts. This parameter is defined by the ratio of the specific elastic energy which has been accumulated in the sample at fracture, calculated from the apparent Young’s modulus and the elastic deformation at fracture, to the whole specific energy expended during the test, determined as the whole area under the stress–strain curve (Fig. 3b). As shown in Table 2, the brittleness param￾eter is the lowest, indicating larger apparent ductility, for the composite containing 40 vol.% of aluminium titanate. This apparent ductility leads to increased energy absorption dur￾ing fracture in the expense of strength, as demonstrated by the fact that only one-fourth of the strength of the A10(A+T) monolith is achieved by these samples. 4.2. Laminates From the above-mentioned results, two laminates with five layers were designed. The system A10A40 was se￾lected to combine low strength and energy absorbing in￾termediate layers of A40(A+T) composition with sufficient strength provided by external layers of A10(A+T) composi￾tion, which was preferred over monophase alumina because of chemical compatibility between the layers. In the system AA10, external and central layers of monophase alumina with high strength were combined with intermediate lay￾ers of A10(A+T) composition, in which microcracks might be developed at most during loading, as discussed above. For symmetry, both laminates were designed with internal central layers of the same composition as that of the corre￾sponding external layer. Considering Young’s modulus and thermal expansion co￾efficients values, given in Table 2, tensile residual stresses are expected in the external and central A10(A+T) and alumina layers in the laminates A10A40 and AA10, re￾spectively. Since residual stresses reduce the strength, thin (∼300
m) intermediate layers and relatively thick (∼1200
m) external and central layers were selected to minimise them. With such structure, negligible residual stresses would be developed in the system AA10 in which high strength values are expected. For the preparation of laminates, suspensions with low viscosity and high solids content result in a better con￾trol of wall thickness formation. Casting rates of 0.53 and 1.31 mm2 min−1 (corresponding to the slopes of the kinet￾ics curves in Fig. 5) are found for alumina and A40(A+T), respectively, which allow a good control of the casting process. The layered green bodies were fabricated by se￾quential slip casting. The thickness of each layer in the laminates was controlled by recalculating the casting time taking into account the sintering shrinkage experienced by the monoliths with the same compositions as those of the layers. Table 3 summarises casting times and the compari￾son between the obtained thickness of the layers, measured directly in the SEM, and the calculated ones. There are sig￾nificant differences (≈25–35%) between the calculated and the obtained widths for the layers made from the A40(A+T) slips, whereas those corresponding to the A10(A+T) slips are inside the uncertainty range associated with the mea￾surement method and the variability between samples. This fact is explained in terms of slip quality, as the one made of A40(A+T) mixture presented the poorest rheological behaviour, thus making the control of casting rate difficult. For the alumina layers, very high casting times (≈1000 s for the central layer, Table 3) are necessary and therefore a possible change in the slip properties could modify the casting kinetics and the obtained thickness. A comparison between the stress–strain behaviour of the sintered laminates and that of monolith samples of the same size is established in Fig. 7. The stress–strain behaviour of the laminate A10A40 was linear up to fracture, such as that of the monoliths that constituted its external and central lay￾ers, made of 10(A+T). This fact indicates that the exter￾nal and central uncracked layers dominate this deformation range. The monolith AA10 behaves also linearly, as both constituents. Nevertheless, the slope of this linear portion was lower for the layered materials than for the large mono￾lithic samples with the same composition as those of the corresponding external layers, due to the effect of the low Young’s modulus internal layers. The stress drop from the failure point of the large mono￾lith samples occurred steeply, and apparent deformation at the maximum load was coincident with that for zero load after fracture (Fig. 7). In the laminates, a step-like way was followed and apparent strain for zero load was larger than that corresponding to failure load. This fracture behaviour is related to the changes in the crack path at the interlay￾ers, as suggested by the fracture surfaces (Fig. 6). The ef￾fect is more significant for A10A40 (Fig. 7b) in which, in addition to the crack deflection due to differences between