Availableonlineatwww.sciencedirect.com CIENCEODIRECT. E≈RS ELSEVIER Journal of the European Ceramic Society 26(2006)2699-2705 www.elsevier.com/locate/jeurcera Piezo-spectroscopic characterization of alumina-aluminium titanate laminates Goffredo de portu a, d. Salvador bueno b. Lorenzo micele a, d Carmen Baudin, Giuseppe Pezzotti, mute of Science and Technology for Ceramics, ISTEC-CNR, Via granarolo, 64-48018 Faenza, Italy b Instituto de Ceramica y vidrio, CSlC-Campus de Cantoblanco, Kelsen 5, 28049 Madrid. e Ceramic Physics Laboratory, Kyoto Institute of Technology KIT, Sakyo-ku, Matsugasaki, 606-8585 Kyoto, Japan Received 19 March 2005: received in revised form 30 June 2005: accepted 8 July 2005 Available online 13 September 2005 Abstract A multilayered alumina-aluminium titanate composite was prepared by a colloidal route from aqueous suspensions. The structure of the laminate was symmetric and constituted of two external Al2O3 layers(width= 1750 um), one central Al2O3 layer(width= 1200 um) and two intermediate thin(width= 315-330 um)Al2O3-Al2TiOs layers Additional monolithic materials with the same compositions as those of the layers were fabricated as reference materials. Youngs modulus of the monoliths was determined by three point bending Dilatometry determinations were performed on green specimens, following the same heating and cooling schedules as those used for sintering the laminate, in order to determine the actual dimensional changes on cooling after sintering. The dimensional changes of the sintered specimens on heating and on cooling were also determined. Microscopic distributions of residual stresses were evaluated by fluorescence piezo-spectroscopy, and they revealed the existence of weak tensile and compressive hydrostatic stresses in the aluminium titanate and alumina layers, respectively. The level and sign of these stresses was in good agreement with those predicted based on analysis of the Young s modulus and the dimensional variations during cooling after sintering of the monoliths with the same compositions as those of the layers. Dimensional variations during cooling after sintering were different from those for sintered materials, which presented hysteresis between heating and cooling. In spite of the presence of compressive residual stresses in layers of the laminate, strength values of notched samples of the laminated specimens were lower than those for monoliths of the same omposition as the external layers. o 2005 Elsevier Ltd. All rights reserved. Keywords: Laminates; Spectroscopy; Thermal expansion; Al2 O3; Al2TiOs 1. Introduction a25-100c=84×10-6°C-1,ac5-100c=9.2×10-6 C-) and aluminium titanate (aa25-1000C=10.9 x lumina(AlO3F-aluminium titanate(Al TiO5)com- 10-6oc-I 100°c=20.5×10-6°C-1,a25-1000c= posites can offer improved flaw tolerance and toughness. -7-27x10-6oC-1) 9 As the toughening mechanisms that Toughening in this system is originated by the residual have been identified in these composites are crack bridging stresses developed, during cooling from the sintering and microcracking, 6, toughening is often associated with mismatch low strength. ai25-1000°C= average exA Laminated materials are being investigated as means to sion coefficient between 25 and 1000oC in the axis combine strength and toughness in ceramic materials. As an example, Russo et al. proposed structures formed by lay Corresponding author. Tel. +3491 7355840: fax: +3491 7355843 ers of equal compositions, made of mixtures of alumina and E-mail address: cbaudin @icv csic es(C. Baudin) aluminium titanate, and dissimilar microstructures In partic 0955-2219/S-see front matter 2005 Elsevier Ltd. All rights reserved. doi: 10.1016/j-jeurceramsoc 2005.07.060
Journal of the European Ceramic Society 26 (2006) 2699–2705 Piezo-spectroscopic characterization of alumina-aluminium titanate laminates Goffredo de Portu a,d, Salvador Bueno b, Lorenzo Micele a,d, Carmen Baudin b,∗, Giuseppe Pezzotti c,d a Institute of Science and Technology for Ceramics, ISTEC-CNR, Via Granarolo, 64-48018 Faenza, Italy b Instituto de Ceramica y Vidrio, CSIC-Campus de Cantoblanco, Kelsen 5, 28049 Madrid, Spain c Ceramic Physics Laboratory, Kyoto Institute of Technology, KIT, Sakyo-ku, Matsugasaki, 606-8585 Kyoto, Japan d Research Institute for Nanoscience, RIN, Sakyo-ku, Matsugasaki, 606-8585 Kyoto, Japan Received 19 March 2005; received in revised form 30 June 2005; accepted 8 July 2005 Available online 13 September 2005 Abstract A multilayered alumina-aluminium titanate composite was prepared by a colloidal route from aqueous suspensions. The structure of the laminate was symmetric and constituted of two external Al2O3 layers (width ∼= 1750m), one central Al2O3 layer (width ∼= 1200m) and two intermediate thin (width ∼= 315–330 m) Al2O3–Al2TiO5 layers. Additional monolithic materials with the same compositions as those of the layers were fabricated as reference materials. Young’s modulus of the monoliths was determined by three point bending. Dilatometry determinations were performed on green specimens, following the same heating and cooling schedules as those used for sintering the laminate, in order to determine the actual dimensional changes on cooling after sintering. The dimensional changes of the sintered specimens on heating and on cooling were also determined. Microscopic distributions of residual stresses were evaluated by fluorescence piezo-spectroscopy, and they revealed the existence of weak tensile and compressive hydrostatic stresses in the aluminium titanate and alumina layers, respectively. The level and sign of these stresses was in good agreement with those predicted based on analysis of the Young’s modulus and the dimensional variations during cooling after sintering of the monoliths with the same compositions as those of the layers. Dimensional variations during cooling after sintering were different from those for sintered materials, which presented hysteresis between heating and cooling. In spite of the presence of compressive residual stresses in the external layers of the laminate, strength values of notched samples of the laminated specimens were lower than those for monoliths of the same composition as the external layers. © 2005 Elsevier Ltd. All rights reserved. Keywords: Laminates; Spectroscopy; Thermal expansion; Al2O3; Al2TiO5 1. Introduction Alumina (Al2O3)–aluminium titanate (Al2TiO5) composites can offer improved flaw tolerance and toughness.1–7 Toughening in this system is originated by the residual stresses developed, during cooling from the sintering temperature, due to the thermal expansion mismatch between alumina (αi25–1000 ◦C = average thermal expansion coefficient between 25 and 1000 ◦C in the axis i, ∗ Corresponding author. Tel.: +34 91 7355840; fax: +34 91 7355843. E-mail address: cbaudin@icv.csic.es (C. Baudin). αa25–1000 ◦C = 8.4 × 10−6 ◦C−1, αc25–1000 ◦C = 9.2 × 10−6 ◦ C−1) 8 and aluminium titanate (αa25–1000 ◦C = 10.9 × 10−6 ◦C−1, αb25–1000 ◦C = 20.5 × 10−6 ◦C−1, αc25–1000 ◦C = −2.7 × 10−6 ◦C−1).9 As the toughening mechanisms that have been identified in these composites are crack bridging and microcracking,6,7 toughening is often associated with low strength. Laminated materials are being investigated as means to combine strength and toughness in ceramic materials.10 As an example, Russo et al.11 proposed structures formed by layers of equal compositions, made of mixtures of alumina and aluminium titanate, and dissimilar microstructures. In partic- 0955-2219/$ – see front matter © 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jeurceramsoc.2005.07.060
2700 G de Porn et al. /Journal of the European Ceramic Society 26(2006)2699-2705 ular, these authors fabricated a laminate with small grained 70 mm x 70 mm x 10 mm dimensions were obtained by slip C1-2 um)and dense external layers with high strength and casting, removed from the moulds and dried in air at room nternal layers with pores and agglomerates. In that system, temperature for at least 24 h. The layered plates, constituted faw tolerance was provided by the R curve behaviour of the by two external and one central alumina layers and two heterogeneous internal layers. The limit of this approach is intermediate composite layers, were fabricated by alternately the difficulty that involves the co-sintering of layers with casting the suspensions. Casting times were fixed to reach the same composition and sufficient microstructural differ- the desired layer thickness considering the casting kinetics ences as to provide significant differences in the mechanical and sintering shrinkage of each suspension. The dried blocks behaviour were sintered in air in an electrical box furnace (Termiber, In a previous work, a laminated structure combining Spain) at heating and cooling rates of 2C min, with 4h external fine grained alumina layers with internal composite dwell at 1200C during heating and 3 h dwell at the maxi- layers of alumina +10 vol % o of aluminium titanate was pro- mum temperature, 1550C Small (12 mm x 5 mm x 5 mm) posed as means to achieve flaw tolerance and high strength. green samples of the monoliths were used for sintering exper- The obtained laminate presented large strain to fracture, as iments in a differential dilatometer(Setaram, Setsys 16/18, compared to that of the monoliths of the same composition of France) with alumina detector. The same sintering schedule the constituent layers. Moreover, under the same loading con- used to obtain the final materials was reproduced in order to ditions, the load drop of the laminate samples, once fracture determine the dimensional variations on cooling was initiated, occurred in step-like form, whereas the mono- The monolithic sintered blocks were machined into bars lith failed catastrophically, thus suggesting the occurrence of 25 mm x 2 mm x 2.5 mm for bending tests(three point of additional energy consuming processes during fracture. bending, 20 mm span, 0.5 mm min-; Microtest, Spain).The Strength values for the laminate samples were slightly lower stress-strain curves were calculated from the load values and (10%) than those corresponding to monoliths of the same the displacement of the central point of the surface of the composition as that of the external layers samples in tension during the tests, and Youngs modulus In this work, the residual stresses developed in this was determined from the linear part of the curves. Reported structure are investigated both experimentally, by piezo- Youngs modulus values are the average of five measurements spectroscopy, and by calculation, using the properties of and errors are the standard deviations. Additional sintered monolithic materials with the same compositions as those samples were machined(10mm x 5 mm x 5 mm) to test in of the layers that constitute the laminate. The effect of the a differential dilatometer(Netzsch, 402 EP, Germany) with residual stresses on the fracture of notched samples is dis- quartz detector, up to 850C and using heating and cooling ussed rates of 5oCmin-I The obtained curves were corrected for the system deformation using a standard of quartz Samples of50mm×6mm×4 mm of the laminates as 2. Experimental well as the monoliths were machined for the mechanical and piezo-spectroscopic measurements For these latter, one 2.1 Processing and thermal and mechanical of the lateral faces(50 mm x 6mm) was diamond polished down to I um and finished with colloidal alumina Mechan- ical tests were performed using the SENVB method(three Monoliths of monophase alumina and alumina +10 vol% point bending, 40 mm span, 0.005 mm min-; Microtest, aluminium titanate and. one laminate were manufactured Spain). The notches were introduced with a depth at about by a colloidal route from aqueous Al2O3 and TiO2 sus- 870 um(a/w=0.4 of the thickness of the external alumina pensions using the optimum green processing conditions layer in the laminate) previously established. 3 and reaction sintering. The struc ture of the laminate was symmetric and constituted of two external Al2O3 layers(width= 1750 um), one central 2.2. Basis of piezo-spectroscopic measurements Al2O3 layer(width 1200 um)and two intermediate thin width 315-330 um)Al2O3-Al2TiO5 layers The stress field distribution along the cross section of The starting powders were commercial a-Al2O3(Condea, the laminated samples was determined by using piezo- HPAO5, USA)and anatase-TiO2(Merck, 808, Germany).A spectroscopy(PS)technique related to the characteristic RI mixture of Al2O3/TiO, with 5 wt% TiO, content was used to R2 doublet produced by chromophoric fluorescence of Cr+ btain the Al203/Al2 TiOs composite materials with 10 voL% impurities in Al203. The principle of relating an observed line Al2TiOs after reaction sintering. The Al2O as well as the shift in a fluorescence spectrum to the state of stress has been Al2O3/TiO2 powders were dispersed in deionised water by described previously by Grabner. When Al2O3 is subjected adding 0.5 wt %(on a dry solids basis)of a carbonic acid to astress o, the change in frequency Av in luminescence line based polyelectrolyte(Dolapix CE64, Zschimmer-Schwarz, is given by the tensorial relationship Germany ). The suspensions (50 vol. solid loading)were ball milled with Al O3 jar and balls during 4h Plates with Av=3lio jj
2700 G. de Portu et al. / Journal of the European Ceramic Society 26 (2006) 2699–2705 ular, these authors fabricated a laminate with small grained (∼=1–2m) and dense external layers with high strength and internal layers with pores and agglomerates. In that system, flaw tolerance was provided by the R curve behaviour of the heterogeneous internal layers. The limit of this approach is the difficulty that involves the co-sintering of layers with the same composition and sufficient microstructural differences as to provide significant differences in the mechanical behaviour. In a previous work,12 a laminated structure combining external fine grained alumina layers with internal composite layers of alumina +10 vol.% of aluminium titanate was proposed as means to achieve flaw tolerance and high strength. The obtained laminate presented large strain to fracture, as compared to that of the monoliths of the same composition of the constituent layers. Moreover, under the same loading conditions, the load drop of the laminate samples, once fracture was initiated, occurred in step-like form, whereas the monolith failed catastrophically, thus suggesting the occurrence of additional energy consuming processes during fracture. Strength values for the laminate samples were slightly lower (∼=10%) than those corresponding to monoliths of the same composition as that of the external layers. In this work, the residual stresses developed in this structure are investigated both experimentally, by piezospectroscopy, and by calculation, using the properties of monolithic materials with the same compositions as those of the layers that constitute the laminate. The effect of the residual stresses on the fracture of notched samples is discussed. 2. Experimental 2.1. Processing and thermal and mechanical characterisation Monoliths of monophase alumina and alumina +10 vol.% aluminium titanate and, one laminate were manufactured by a colloidal route from aqueous Al2O3 and TiO2 suspensions using the optimum green processing conditions previously established12,13 and reaction sintering. The structure of the laminate was symmetric and constituted of two external Al2O3 layers (width ∼= 1750m), one central Al2O3 layer (width ∼= 1200m) and two intermediate thin (width ∼= 315–330m) Al2O3–Al2TiO5 layers. The starting powders were commercial -Al2O3 (Condea, HPA05, USA) and anatase-TiO2 (Merck, 808, Germany). A mixture of Al2O3/TiO2 with 5 wt.% TiO2 content was used to obtain the Al2O3/Al2TiO5 composite materials with 10 vol.% Al2TiO5 after reaction sintering. The Al2O3 as well as the Al2O3/TiO2 powders were dispersed in deionised water by adding 0.5 wt.% (on a dry solids basis) of a carbonic acid based polyelectrolyte (Dolapix CE64, Zschimmer-Schwarz, Germany). The suspensions (50 vol.% solid loading) were ball milled with Al2O3 jar and balls during 4h. Plates with 70 mm × 70 mm × 10 mm dimensions were obtained by slip casting, removed from the moulds and dried in air at room temperature for at least 24 h. The layered plates, constituted by two external and one central alumina layers and two intermediate composite layers, were fabricated by alternately casting the suspensions. Casting times were fixed to reach the desired layer thickness considering the casting kinetics and sintering shrinkage of each suspension. The dried blocks were sintered in air in an electrical box furnace (Termiber, Spain) at heating and cooling rates of 2 ◦C min−1, with 4 h dwell at 1200 ◦C during heating and 3 h dwell at the maximum temperature, 1550 ◦C. Small (12 mm × 5 mm × 5 mm) green samples of the monoliths were used for sintering experiments in a differential dilatometer (Setaram, Setsys 16/18, France) with alumina detector. The same sintering schedule used to obtain the final materials was reproduced in order to determine the dimensional variations on cooling. The monolithic sintered blocks were machined into bars of 25 mm × 2 mm × 2.5 mm for bending tests (three point bending, 20 mm span, 0.5 mm min−1; Microtest, Spain). The stress-strain curves were calculated from the load values and the displacement of the central point of the surface of the samples in tension during the tests, and Young’s modulus was determined from the linear part of the curves. Reported Young’s modulus values are the average of five measurements and errors are the standard deviations. Additional sintered samples were machined (10 mm × 5 mm × 5 mm) to test in a differential dilatometer (Netzsch, 402 EP, Germany) with quartz detector, up to 850 ◦C and using heating and cooling rates of 5 ◦C min−1. The obtained curves were corrected for the system deformation using a standard of quartz. Samples of 50 mm × 6 mm × 4 mm of the laminates as well as the monoliths were machined for the mechanical and piezo-spectroscopic measurements. For these latter, one of the lateral faces (50 mm × 6 mm) was diamond polished down to 1m and finished with colloidal alumina. Mechanical tests were performed using the SENVB method (three point bending, 40 mm span, 0.005 mm min−1; Microtest, Spain). The notches were introduced with a depth at about 870m (a/w ∼= 0.4 of the thickness of the external alumina layer in the laminate). 2.2. Basis of piezo-spectroscopic measurements The stress field distribution along the cross section of the laminated samples was determined by using piezospectroscopy (PS) technique related to the characteristic R1, R2 doublet produced by chromophoric fluorescence of Cr3+ impurities in Al2O3. The principle of relating an observed line shift in a fluorescence spectrum to the state of stress has been described previously by Grabner.14 When Al2O3 is subjected to a stress σ, the change in frequency ν in luminescence line is given by the tensorial relationship: ∆ν = 1 3Πiiσjj (1)
G de Portu et al. Joumal of the European Ceramic Sociery 26(2006)2699-2705 where Ii is referred to as the PS coefficient (i.e, relating tra were analysed with a commercial software LabSpec 4.02, frequency to stress)and oj is the first invariant of the stress Horiba/Jobin-Ivon). The frequency shifts were obtained by tensor(oj /3 being normally referred to as the mean normal subtracting from the centre of the peak recorded under stress, stress). the centre frequency of the peak obtained in the unstressed In principle, the overall residual stress field is due to differ- state ent thermal expansion and elastic mismatch between the con- Microscopic stress distributions were measured by col stituent ceramic phases of the Al2O3/Al]TiOs laminate and lecting linear profiles of spectra on the specimen cross- consists of two separate components: (i)a microscopic stress sections. The automatically collected profiles of spectra were field deriving from the microstructural scale from grain-to- 10 um-spaced. Specimens were placed on a mapping device ain thermal and elastic mismatches between Al2O3 and (lateral resolution of 0. 1 um), which was connected to a per- Al2T1O5 phases; and (ii)a macroscopic stress field, which sonal computer to drive highly precise displacements(along established to fulfil equilibrium conditions between adja- both X and y axes)on the specimen surface cent layers. To measure merely the second contribution of Under the assumption that composites prepared accord- the stress, data were analyzed following the approach sug- ing to the above mentioned process are polycrystalline and gested by De Portu et al. for monolithic composites with without any significant texture, the average component of ence of the ratio matrix/secondary-phase in the determined by rearranging Eq (1)as follows 2+033)/3 is calculated no laminated structure. These authors established the influ- the hydrostatic stress, (o)=(o1+o zero-stress position of the RI peak as well as of the piezo- △1 spectroscopic coefficient. Moreover, they demonstrated that (a) the average uniaxial PS coefficient, IuNi, characterizing 3()UN the linear dependence of the peak shift on stress, strongly where( IT)UNI represents the PS coefficient that is determined depends on many factors specific of the material, especially from a calibration where the stress is uniaxial. It is important in processing derived parameters such as grain size, presence to emphasize that Eq(2) provides the average hydrostatic of other phases, porosity, etc. Hence, in order to obtain a flu- stress that is experienced by the phase that generates the spe kes into account the effect of the microstructure on the stress in laminated structures, far away from external edges, as orescence spectra and piezo-spectroscopic coefficient, which troscopic signal for which the shift Av is monitored. In fact, field before lamination, a preliminary calibration procedur residual stress field can be considered to be of a biaxial nature is required case by case for determining the /IuNI value perti- accordingly, it should be computed as 2/3 oi. On the other nent to each material. For this reason, the frequencies used as hand, nearby the edges the residual stress field is typically standard value for the stress-free material were collected three-dimensional, thus requiring the use of a mean normal on monolithic AlO and composite Al2O3/Al2TiOs bars stress oj in Eq- (1). For stress measurement, RI band was prepared following the same processing as that of the multi- used in order to avoid errors in fitting due to the superpo- layered specimens. In this way the RI peak of each stress free sition of a line of reference Hg/Ne lamp spectrum with R2 material can be precisely obtained and compared with that band. In relation to the weak deviation from linearity of rI of the same composition in the laminated structure More- calibration line evidenced by Maet al., 6 considered the high ver, calibration of spectral shift as a function of externally correlation coefficient obtained by lines referred to laminated applied load for each composition was done, according to material and the low stress measured, it was supposed to be the procedures described in Section 3. 1. Using Eq (1)the Ps negligible oefficient Iii for both pure alumina and, for the first time for alumina-aluminium titanate composites were obtained. 3. Results and discussion 2.3. Determination of the residual stresses 3.1. Determination of the residual stresses To collect fluorescence spectra, the spectrometer ratus(ISA, T 64000 Jovin-Yvon) employed in the pre Unstressed peak position was obtained acquiring an array experiments used, as excitation source, an argon-ion of 100 spectra on the surface of unstressed reference bars operating at a wavelength of 488 nm with a power of 400 mw. and averaging all the values of the peak centre. For the eval For obtaining micron-scale magnification, an optical micro- uation of PS coefficient, bending bars of both composites scope lens was used both to focus the laser on the sample and were mounted on a four-point bending jig and loaded with a to collect the scattered signal. Scattered frequencies were an known load below fracture stress; after the load was applied, lyzed with a triple monochromator equipped with a charge the whole jig was moved under the microscope and spectra coupled device( CCD)camera. When focussed by an optical recorded every 40 um from the compressive towards the ten- microscope, the dimension of the laser spot on the samples sile side of the specimen. To reduce the scattering of data, was 5 um (i.e, using a x20 optical lens). Thermal and instru- and consequently improve PS coefficient determination, the mental fluctuations were compensated by monitoring the calibration was repeated three times and the values obtained pectrum using a Hg/Ne discharge lamp. The recorded spec- were averaged. The load was then converted to stress, o, using
G. de Portu et al. / Journal of the European Ceramic Society 26 (2006) 2699–2705 2701 where Πii is referred to as the PS coefficient (i.e., relating frequency to stress) and σjj is the first invariant of the stress tensor (σjj/3 being normally referred to as the mean normal stress). In principle, the overall residual stress field is due to different thermal expansion and elastic mismatch between the constituent ceramic phases of the Al2O3/Al2TiO5 laminate and consists of two separate components: (i) a microscopic stress field deriving from the microstructural scale from grain-tograin thermal and elastic mismatches between Al2O3 and Al2TiO5 phases; and (ii) a macroscopic stress field, which is established to fulfil equilibrium conditions between adjacent layers. To measure merely the second contribution of the stress, data were analyzed following the approach suggested by De Portu et al.15 for monolithic composites with no laminated structure. These authors established the influence of the ratio matrix/secondary-phase in the determined zero-stress position of the R1 peak as well as of the piezospectroscopic coefficient. Moreover, they demonstrated that the average uniaxial PS coefficient, ΠUNI, characterizing the linear dependence of the peak shift on stress, strongly depends on many factors specific of the material, especially in processing derived parameters such as grain size, presence of other phases, porosity, etc. Hence, in order to obtain a fluorescence spectra and piezo-spectroscopic coefficient, which takes into account the effect of the microstructure on the stress field before lamination, a preliminary calibration procedure is required case by case for determining the ΠUNI value pertinent to each material. For this reason, the frequencies used as a standard value for the stress-free material were collected on monolithic Al2O3 and composite Al2O3/Al2TiO5 bars prepared following the same processing as that of the multilayered specimens. In this way the R1 peak of each stress free material can be precisely obtained and compared with that of the same composition in the laminated structure. Moreover, calibration of spectral shift as a function of externally applied load for each composition was done, according to the procedures described in Section 3.1. Using Eq. (1) the PS coefficient Πii for both pure alumina and, for the first time, for alumina-aluminium titanate composites were obtained. 2.3. Determination of the residual stresses To collect fluorescence spectra, the spectrometer apparatus (ISA, T 64000 Jovin-Yvon) employed in the present experiments used, as excitation source, an argon-ion laser operating at a wavelength of 488 nm with a power of 400 mW. For obtaining micron-scale magnification, an optical microscope lens was used both to focus the laser on the sample and to collect the scattered signal. Scattered frequencies were analyzed with a triple monochromator equipped with a charge coupled device (CCD) camera. When focussed by an optical microscope, the dimension of the laser spot on the samples was 5m (i.e., using a ×20 optical lens). Thermal and instrumental fluctuations were compensated by monitoring the spectrum using a Hg/Ne discharge lamp. The recorded spectra were analysed with a commercial software (LabSpec 4.02, Horiba/Jobin-Ivon). The frequency shifts were obtained by subtracting from the centre of the peak recorded under stress, the centre frequency of the peak obtained in the unstressed state. Microscopic stress distributions were measured by collecting linear profiles of spectra on the specimen crosssections. The automatically collected profiles of spectra were 10m-spaced. Specimens were placed on a mapping device (lateral resolution of 0.1m), which was connected to a personal computer to drive highly precise displacements (along both X and Y axes) on the specimen surface. Under the assumption that composites prepared according to the above mentioned process are polycrystalline and without any significant texture, the average component of the hydrostatic stress, σ=(σ11 + σ22 + σ33)/3, is calculated by rearranging Eq. (1) as follows:16 σ = ∆ν 3ΠUNI (2) where ΠUNI represents the PS coefficient that is determined from a calibration where the stress is uniaxial. It is important to emphasize that Eq. (2) provides the average hydrostatic stress that is experienced by the phase that generates the spectroscopic signal for which the shift ν is monitored. In fact, in laminated structures, far away from external edges,17 the residual stress field can be considered to be of a biaxial nature; accordingly, it should be computed as 2/3 σij. On the other hand, nearby the edges the residual stress field is typically three-dimensional, thus requiring the use of a mean normal stress σjj in Eq.(1). For stress measurement, R1 band was used in order to avoid errors in fitting due to the superposition of a line of reference Hg/Ne lamp spectrum with R2 band. In relation to the weak deviation from linearity of R1 calibration line evidenced by Ma et al.,16 considered the high correlation coefficient obtained by lines referred to laminated material and the low stress measured, it was supposed to be negligible. 3. Results and discussion 3.1. Determination of the residual stresses Unstressed peak position was obtained acquiring an array of 100 spectra on the surface of unstressed reference bars and averaging all the values of the peak centre. For the evaluation of PS coefficient, bending bars of both composites were mounted on a four-point bending jig and loaded with a known load below fracture stress; after the load was applied, the whole jig was moved under the microscope and spectra recorded every 40 m from the compressive towards the tensile side of the specimen. To reduce the scattering of data, and consequently improve PS coefficient determination, the calibration was repeated three times and the values obtained were averaged. The load was then converted to stress, σ, using
G. de Portu et al. Journal of the European Ceramic Society 26(2006)2699-2705 0.0 Al2O3/Al2TOs Fig. 2. Structure of the laminate: two thick(1750 um)external AlzO3 layers, observed partially in the figure, one central Al2 O3 layer(1200 um)and two intermediate thin(315-330 um)Al2O3-Al2TiOs layers Scanning electron micrograph of a polished surface(thermally etched, 1500 C-2 min): Al2O3 (dark grey)Al2 TiOs(clear grey). 0.6 Induced stress / GPa Fig 1. Stress dependence of the position of RI band in the materials analyzed 10002000300040005000 in this work: Al2O3(top) and Al2O3/Al]TiOs composite(bottom) the standard four-point-bending elastic equation and the pea shift, Av, plotted as a function of the applied stress(Fig. 1) length along the corss section /um The average IuNI was obtained from the slope of the straight line o versus Av. Table I summarizes the values of salient Fig 3. Profiles of macroscopic residual stress along the cross section. Pure piezo-spectroscopic characteristics of the materials studied in Al2O3 layers were in compression, whereas Al2O3/Al TiOs composite lay- the present investigation. Significant diversity was evidenced ers underwent tensile stresses. The error for residual stress value was #2 MPa in data scattering between the pure alumina and the alumina- for the alumina layers and #4 MPa tor the composite layers aluminum titanate composite: in the latter, data were more spread and, consequently, the correlation between peak shift 2 shows the structure of the obtained symmetric lam- and induced stress was lower. The higher scattering in the inate. It was constituted of two thick(1750 um) external omposite can be attributed to the effect of residual stresses Al2O3 layers and one central Al2O3 layer(1200 um)and at the microscopic level on the stress state of the specimen two intermediate thin(315-330 um)Al2O3-Al2TiOs layers (Section 2.2). In fact, as Al2TiO5 is an extremely anisotropic he oscopic residual stress profile along the cross phase in terms of thermal expansion, microscopic residual section determined using Eq (2)is plotted in Fig. 3. Results stresses are expected in the Al2O3/Al2 TiO5 composites, as are the average of five profiles collected along the cross discussed in the introduction. The level and, moreover, the section to reduce the scattering of data and to verify the uni- sign of such stresses would depend on the particular grain to formity of residual stress field. In this way, the errors relative grain orientation, leading to the high scatter observed (Fig. 1, to stress calculation were +2 and +4 MPa for the AlzO3 and Table 1). Al2O3/Al2TiO5 layers, respectively. According this profile. troscopic characteristics of RI band of the investigated materials Materal Iuni(cm -/GPa) Hydro(cm/GPa) 2.73±0.03 8.19±0.09 Al O3/AlTOS 2.2±0.1 1440700±0.01 Confidence intervals of the PS coefficient were calculated with the software Origin 6.0, with a probability of 95%
2702 G. de Portu et al. / Journal of the European Ceramic Society 26 (2006) 2699–2705 Fig. 1. Stress dependence of the position of R1 band in the materials analyzed in this work: Al2O3 (top) and Al2O3/Al2TiO5 composite (bottom). the standard four-point-bending elastic equation and the peak shift, ν, plotted as a function of the applied stress (Fig. 1). The average ΠUNI was obtained from the slope of the straight line σ versus ν. Table 1 summarizes the values of salient piezo-spectroscopic characteristics of the materials studied in the present investigation. Significant diversity was evidenced in data scattering between the pure alumina and the aluminaaluminum titanate composite: in the latter, data were more spread and, consequently, the correlation between peak shift and induced stress was lower. The higher scattering in the composite can be attributed to the effect of residual stresses at the microscopic level on the stress state of the specimen (Section 2.2). In fact, as Al2TiO5 is an extremely anisotropic phase in terms of thermal expansion,9 microscopic residual stresses are expected in the Al2O3/Al2TiO5 composites, as discussed in the introduction. The level and, moreover, the sign of such stresses would depend on the particular grain to grain orientation, leading to the high scatter observed (Fig. 1, Table 1). Fig. 2. Structure of the laminate: two thick (1750 m) external Al2O3 layers, observed partially in the figure, one central Al2O3 layer (1200 m) and two intermediate thin (315–330 m) Al2O3–Al2TiO5 layers. Scanning electron micrograph of a polished surface (thermally etched, 1500 ◦C−2 min): Al2O3 (dark grey) Al2O3-Al2TiO5 (clear grey). Fig. 3. Profiles of macroscopic residual stress along the cross section. Pure Al2O3 layers were in compression, whereas Al2O3/Al2TiO5 composite layers underwent tensile stresses. The error for residual stress value was±2 MPa for the alumina layers and ±4 MPa for the composite layers. Fig. 2 shows the structure of the obtained symmetric laminate. It was constituted of two thick (1750 m) external Al2O3 layers and one central Al2O3 layer (1200m) and two intermediate thin (315–330m) Al2O3–Al2TiO5 layers. The macroscopic residual stress profile along the cross section determined using Eq. (2) is plotted in Fig. 3. Results are the average of five profiles collected along the cross section to reduce the scattering of data and to verify the uniformity of residual stress field. In this way, the errors relative to stress calculation were ±2 and ±4 MPa for the Al2O3 and Al2O3/Al2TiO5 layers, respectively. According this profile, Table 1 Salient piezo-spectroscopic characteristics of R1 band of the investigated materials Material Πuni (cm−1/GPa) Πhydro (cm−1/GPa) R2 Unstressed peak position (cm−1) Al2O3 2.73 ± 0.03 8.19 ± 0.09 0.993 14407.31 ± 0.01 Al2O3/Al2TiO5 2.2 ± 0.1 6.6 ± 0.3 0.90 14407.00 ± 0.01 Confidence intervals of the PS coefficient were calculated with the software Origin 6.0, with a probability of 95%.
G de Portu et al. Journal of the European Ceramic Sociery 26(2006)2699-2705 2703 the Al2O3 layers underwent a constant compressive stress whereas they present an hystheresis from about 700C for of about 20 MPa, whereas the residual stress distribution in the composite(Fig. 4b), similar to that reported elsewhere the Al2O3/Al2 TiOs layers, tensile on the whole, varied con- for aluminium titanate based materials. The presence of siderably through the layer thickness: it was approximately the hysteresis in the curve corresponding to the composite, 5 MPa in the centre of the layers and rose up to 20 MPa at the does not allow the evaluation of the residual stresses from the Interfaces standard thermal expansion analysis on heating. In order to evaluate the actual differences between the 3. 2. Properties of the monolith and residual stresses dimensional variations of the different layers during cool- ing, the corresponding part of the sintering curves shown in In a previous work, we determined the thermal expan ig. 5a were analysed. In alumina based materials, defor- ion of the pure Al2O3 and the Al2O3/Al2 TiOs sin mation mismatch at temperatures higher than 1200C can tered monoliths during heating, which were coincident be accomodated by diffusion, 22 whereas, from 1200. to with those determined here(Fig. 4). From Fig. 4,aver room temperature, this mismatch originates stresses. There- age thermal expansion values on heating of 8.2+0.1 and fore, in order to evaluate the stress level and sign in the 78±0.1×10-6°C- are derived for pure Al2O3 and the laminate. the differences between the dimensional variations composite, respectively. If these data are used to evaluate the of specimens during cooling from the"stress free"tempera residual stress state in the laminate, the expected sign of the ture, 1200C, were analysed In Fig 5b, these variations are residual stresses would be the opposite of that experimentally plotted toghether for the two monoliths, after correction for of temperatures(1200-50 C)the shrinkage for the compo than the composite layers and thus, the former layers would ite (10. x 10-C )is slightly larger than that of Al2 O3 be in tension in the laminate. 9,20 This apparent contradic (9.8x 10C ), which would imply compressive and tion is clarified looking at the cooling part of the dilatometry tensile residual stresses in the alumina and in the compos- curves in Fig 4. They are coincident for Al2O3(Fig. 4a) 0,00 0.008 -0,02 A10 0007 -0,12 0,003 -0.14 02004006008001000120014001600 0.000 TrC] TrO 0008 0.0005 0.007 0005 0.004 00020 0.003 -0.0025 020040060080010001200 001 0000 Trc] length variation)f TrcI ated following the A:Al2O3;,A10:Al2O3+10 2TIOs. (a)Complete t cycle.(b) Fig 4. Dilatometric curves(AL/Lo=length variation) on heating and cool- Cooling from 1200oC. after at1200°C ng of the sintered monoliths: (a)Al2O3; ( b)AlO3+10 vol. Al] TiO
G. de Portu et al. / Journal of the European Ceramic Society 26 (2006) 2699–2705 2703 the Al2O3 layers underwent a constant compressive stress of about 20 MPa, whereas the residual stress distribution in the Al2O3/Al2TiO5 layers, tensile on the whole, varied considerably through the layer thickness: it was approximately 5 MPa in the centre of the layers and rose up to 20 MPa at the interfaces. 3.2. Properties of the monolith and residual stresses In a previous work,18 we determined the thermal expansion of the pure Al2O3 and the Al2O3/Al2TiO5 sintered monoliths during heating, which were coincident with those determined here (Fig. 4). From Fig. 4, average thermal expansion values on heating of 8.2 ± 0.1 and 7.8 ± 0.1 × 10−6 ◦C−1 are derived for pure Al2O3 and the composite, respectively. If these data are used to evaluate the residual stress state in the laminate, the expected sign of the residual stresses would be the opposite of that experimentally determined here, as the layers of pure Al2O3, with larger thermal expansion coefficient, would contract more on cooling than the composite layers and thus, the former layers would be in tension in the laminate.19,20 This apparent contradiction is clarified looking at the cooling part of the dilatometry curves in Fig. 4. They are coincident for Al2O3 (Fig. 4a) Fig. 4. Dilatometric curves (L/L0 = length variation) on heating and cooling of the sintered monoliths: (a) Al2O3; (b) Al2O3 + 10 vol.% Al2TiO5. whereas they present an hystheresis from about 700 ◦C for the composite (Fig. 4b), similar to that reported elsewhere for aluminium titanate based materials.21 The presence of the hysteresis in the curve corresponding to the composite, does not allow the evaluation of the residual stresses from the standard thermal expansion analysis on heating. In order to evaluate the actual differences between the dimensional variations of the different layers during cooling, the corresponding part of the sintering curves shown in Fig. 5a were analysed. In alumina based materials, deformation mismatch at temperatures higher than 1200 ◦C can be accomodated by diffusion,22 whereas, from 1200 ◦C to room temperature, this mismatch originates stresses. Therefore, in order to evaluate the stress level and sign in the laminate, the differences between the dimensional variations of specimens during cooling from the “stress free” temperature, 1200 ◦C, were analysed. In Fig. 5b, these variations are plotted toghether for the two monoliths, after correction for coincident dimensions at 1200 ◦C. In the considered interval of temperatures (1200–50 ◦C) the shrinkage for the composite (∼=10.1 × 10−6 ◦C−1) is slightly larger than that of Al2O3 (∼=9.8 × 10−6 ◦C−1), which would imply compressive and tensile residual stresses in the alumina and in the composFig. 5. Dilatometric curves (L/L0 = length variation) for green compacts of the two studied monoliths heat treated following the sintering schedule. A: Al2O3; A10: Al2O3 + 10 vol.% Al2TiO5. (a) Complete thermal cycle. (b) Cooling from 1200 ◦C, after correction for coincident dimensions at 1200 ◦C and apparatus deformation.
G de Port et al. /Journal of the European Ceramic Society 26(2006)2699-2705 ite layers, respectively, as determined by piezo-spectroscopy reduces the overall residual stress in the superficial region (Fig. 3) investigated by Appart from differences in the dimensional variations of The notch entered 870 Hm in both, the laminate and the the different layers during cooling from the sintering temper- monolith, which corresponded to =0. 4 the thickness of the ature, the residual stresses in a laminate are determined by the first layer in the laminate. Strength values of the notched sar elastic properties of the constituent layers. Youngs modulus ples calculated using the approach by Timoshenko for stress of composite materials are usually dependent on the method through a laminated structure 24 were 91+3 MPa for the lam- used for the determination. In general, the stiffer phase domi- inate and 104+5 MPa for the monolith, thus, following the nates over the obtained value when dynamic methods are used same trend (10% lower for the laminate) as that previously whereas static methods are more sensitive to grain bound- observed in un-notched samples. 2 This strength decrease in ary characteristics and microscopic residual stresses. These the laminate, in spite of the compressive residual stresses differences are due to the characteristics of the deforma- existing in the external alumina layer of the laminate, where tions involved in each case, small and instantaneous in the the whole notch was located might be related with the lower dynamic case and larger and distributed in time in the static stiffness of the laminate as compared to the pure Al2O3 mono- case From Fig 5b and the discussion above, the expected lith, due to the lower modulus of the composite layers and erence between the dimensional variations of the layers the microstructural characteristics of the interfaces between cooling is 2 x 10-4 in the studied laminate, which is of hemically incompatible layers the same order as strains involved in bending tests(10-4) e values of static Youngs modulus deter mined in bending were chosen in this work. Young's modulus 4. Conclusions for the pure Al2O3 monolith, 376+6GPa, was larger than that for the Al2O3-Al2T1O5, 272+10 GPa. The fact that It has been possible to determine the microscopic distri- residual stresses decreased at the centre of the composite butions of residual stresses developed in a laminate, consti- layers, which has also been observed in Al2O3-ZrO2 lam- tuted by thick external(=1750 um)and central(=1200 um) inated composites,2 can be attributed to the lower elastic alumina layers and thin(320 um) intermediate alumina modulus of Al2O3-Al2T1Os compared to pure Al2O3 +10 vol %o aluminium titanate layers, by fluorescence piezo- spectroscopy. Weak tensile and compressive hydrostatic The level and sign of the expected residual stresses stresses in the aluminium titanate and alumina layers, respec within symmetrical laminate can be evaluated using simpl fied model of symmetric plate constituted by alternate layers tively, have been detected. The level and sign of these stresses of the same thickness having a uniform biaxial distribution of is in good agreement with those expected from the Youngs modulus and the actual dimensional variations experienced residual stresses at the centre of the pure alo a and the by the constituent layers on cooling after sintering composite layers, A10, are given by OA= 1+(EAnAhA/EAo AlohAlo) Work supported in part by the European Communitys A10=0A Human Potential Programme under contract HPRN-CT- 2002-00203 SICMAC]. The Spanish authors acknowl- where Ae is the thermal expansion mismatch between the edge the support of the projects CICYT MAT 2003-00836 layers, nAAlo and hA Alo are the number and thickness of the BPD2001-1(Spain ).G.de Portu is grateful to Japan Society for the Promotion of Science (SPS)and Italian National where EAAlo is the Youngs modulus and v is the poissons Research Council(CNR) for the financial support to his work ratio.Assuming the same width for the three Al203 layers in Japan. The contribution of Italian Ministry for Foreign as that of the central one in the laminate studied here and using for Ae the value derived from the cooling part of the Affair(MAE), which supported the creation of rin is also sintering curves(2 x 10+, Fig 5b)compressive stress at the gratefully acknowledged centre of the central alumina layer of about 9 MPa and ten- sile stresses of about 48 MPa at the centres of the composite References layers, which are in reasonable agreement with the experi- mental values(Fig 3), are obtained. Differences would be in 1. Runyan, J. L. and Bennison, S. J, Fabrication of flaw-tolerant part due to the"edge effect"discussed elsewhere 7 briefly, aluminum-titanate-reinforced alumina. J. Eur Ceram. Soc. 1991. 7 near edges the residual stress state is not biaxial, because the edges themselves must be traction free. For that reason, the 2. Padture. N. P. Bennison. S. and Chan. H. M. Flaw-tolerance and crack-resistance properties of alumina-aluminium titanate com- actual stress state is the superimposition of the bulk residual osites with tailored microstructures. J. Am. Ceram. Soc.. 1993. 76 stress and another component opposite to the first one, which 2312-2320
2704 G. de Portu et al. / Journal of the European Ceramic Society 26 (2006) 2699–2705 ite layers, respectively, as determined by piezo-spectroscopy (Fig. 3). Appart from differences in the dimensional variations of the different layers during cooling from the sintering temperature, the residual stresses in a laminate are determined by the elastic properties of the constituent layers. Young’s modulus of composite materials are usually dependent on the method used for the determination. In general, the stiffer phase dominates over the obtained value when dynamic methods are used whereas static methods are more sensitive to grain boundary characteristics and microscopic residual stresses. These differences are due to the characteristics of the deformations involved in each case, small and instantaneous in the dynamic case and larger and distributed in time in the static case. From Fig. 5b and the discussion above, the expected difference between the dimensional variations of the layers on cooling is 2 × 10−4 in the studied laminate, which is of the same order as strains involved in bending tests (∼10−4). Consequently, the values of static Young’s modulus determined in bending were chosen in this work. Young’s modulus for the pure Al2O3 monolith, 376 ± 6 GPa, was larger than that for the Al2O3–Al2TiO5, 272 ± 10 GPa. The fact that residual stresses decreased at the centre of the composite layers, which has also been observed in Al2O3–ZrO2 laminated composites15,23 can be attributed to the lower elastic modulus of Al2O3–Al2TiO5 compared to pure Al2O3. The level and sign of the expected residual stresses within symmetrical laminate can be evaluated using simpli- fied model of symmetric plate constituted by alternate layers of the same thickness having a uniform biaxial distribution of stresses across each layer.19 Using this approach, the arising residual stresses at the centre of the pure Al2O3, A, and the composite layers, A10, are given by: σA = − ∆εE A 1 + (E AnAhA/E A10nA10hA10) (3) σA10 = −σA nA nA10 hA hA10 (4) where ε is the thermal expansion mismatch between the layers, nA,A10 and hA,A10 are the number and thickness of the layers, respectively, and E A,A10 is equal to EA,A10/(1 − ν) where EA,A10 is the Young’s modulus and ν is the Poisson’s ratio. Assuming the same width for the three Al2O3 layers as that of the central one in the laminate studied here and using for ε the value derived from the cooling part of the sintering curves (2 × 10−4, Fig. 5b) compressive stress at the centre of the central alumina layer of about 9 MPa and tensile stresses of about 48 MPa at the centres of the composite layers, which are in reasonable agreement with the experimental values (Fig. 3), are obtained. Differences would be in part due to the “edge effect” discussed elsewhere:17 briefly, near edges the residual stress state is not biaxial, because the edges themselves must be traction free. For that reason, the actual stress state is the superimposition of the bulk residual stress and another component opposite to the first one, which reduces the overall residual stress in the superficial region investigated by piezo-spectroscopy. The notch entered 870m in both, the laminate and the monolith, which corresponded to ∼=0.4 the thickness of the first layer in the laminate. Strength values of the notched samples calculated using the approach by Timoshenko for stress through a laminated structure24 were 91±3 MPa for the laminate and 104±5 MPa for the monolith, thus, following the same trend (∼=10% lower for the laminate) as that previously observed in un-notched samples.12 This strength decrease in the laminate, in spite of the compressive residual stresses existing in the external alumina layer of the laminate, where the whole notch was located, might be related with the lower stiffness of the laminate as compared to the pure Al2O3 monolith, due to the lower modulus of the composite layers and the microstructural characteristics of the interfaces between chemically incompatible layers.18 4. Conclusions It has been possible to determine the microscopic distributions of residual stresses developed in a laminate, constituted by thick external (∼=1750m) and central (∼=1200m) alumina layers and thin (∼=320m) intermediate alumina +10 vol.% aluminium titanate layers, by fluorescence piezospectroscopy. Weak tensile and compressive hydrostatic stresses in the aluminium titanate and alumina layers, respectively, have been detected. The level and sign of these stresses is in good agreement with those expected from the Young’s modulus and the actual dimensional variations experienced by the constituent layers on cooling after sintering. Acknowledgements Work supported in part by the European Community’s Human Potential Programme under contract HPRN-CT- 2002-00203 [SICMAC]. The Spanish authors acknowledge the support of the projects CICYT MAT 2003-00836 and CAM, GRMAT0707-2004 and the grant CSIC I3PBPD2001-1 (Spain). G. de Portu is grateful to Japan Society for the Promotion of Science (JSPS) and Italian National Research Council (CNR) for the financial support to his work in Japan. The contribution of Italian Ministry for Foreign Affair (MAE), which supported the creation of RIN is also gratefully acknowledged. References 1. Runyan, J. L. and Bennison, S. J., Fabrication of flaw-tolerant aluminum-titanate-reinforced alumina. J. Eur. Ceram. Soc., 1991, 7, 93–99. 2. Padture, N. P., Bennison, S. J. and Chan, H. M., Flaw-tolerance and crack-resistance properties of alumina-aluminium titanate composites with tailored microstructures. J. Am. Ceram. Soc., 1993, 76, 2312–2320.
G. de Portu et al. Journal of the European Ceramic Society 26(2006)2699-2705 2705 3. Bartolome, J, Requena, J, Moya, J.S.,Li, M. and Guiu, F, Cyclic 15. De Portu. G. Micele. L. Sekiguchi. Y and Pezzotti G. Measurement fatigue crack growth resistance of AlzO3-AlTiOs composites Acta f residual stress distributions in AlO3/3Y-TZP multilayered com- fater,1996,44.1361-1370 sites by fluorescence and Raman microprobe piezo-spectroscopy 4. Bartolome, J, Requena, J, Moya, J.S., Li, M. and Guiu, F, Cyclic Acta Mater. 2005. 53. 1511-1520 fatigue of Al2O3-AlTiOs composites in direct push-pull. Fatigu 16. Ma, Q. and Clarke, D measurement in single-crys Fract. Eng. Mater. Struct, 1997, 20, 789-798 d polycrystalline cerar their optical fluorescence.J.Am 5. Uribe. R. and Baud Ceram Soc.,1993,76.14 titanate particles of controlled size on the thermal shock resistance 17. Sergo, V, Lipkin, D. M, De Portu, G. and Clarke, D. R of alumina. J. Am. Ceram. Soc. 200 46-850. tresses in alumina/zirconia laminates. J. Am. Ceram. Soc.. 19 197 80 6. Lawn, B. R. Padture. N. P. Braun. L. M. and Bennison, S J. Model 1633-1638 for toughness curves in two-phase ceramics: I, basic fracture mechan- 18. Bueno, S. and Baudin, C, In situ developed laminates with large ics.J.Am. Ceran.Soc,1993,76,2235-2240 microstructural differences between layers of similar composition. J. 7. Padture. N. P an. J. L. Bennison, S. J, Braun, L. M. and Mater. Sci., in pr awn, B.R. Model for toughness curves in two-phase ceramics: Il, 19. Oel, H. J. and Frechette, V. D, Stress Distribution in mult microstructural vanables. J. A. Ceram. Soc. 1993. 76, 2241-2247 mposites with planar interfaces. J. Am. Ceram. 8. Taylor. D, Themal expansion data: Ill. Sesquioxides, M2O3 with the 1967,50.542-549 corundum and the A-, B- and C-M2O3. Br. Ceram. Trans. J, 1984 20. Ho, S, Hillman, C, Lange, F. F. and Suo, Z, Surface cracking in under biaxial, residual compressive stress. J. Am. Ceram. Soc. 9. Taylor, D. Thermal expansion data. XI. Complex oxides, A2BOs 995,78,2353-2359 21. Hasselman, D. P. H. Donaldson, K. Y, Anderson, E M. and Johnson T. A, Effect of thermal history on the thermal diffusivity and thermal Annu. Rev. Mater Sci. 1997. 27. 249-282 expansion of an alumina-aluminium titanate composite. J Am Ceram 11. Russo. C.J. Harmer, M. P, Chan. H. M. and Miller. G. A, Design of Soc.,1993,76 a laminated ceramic composite for improved strength and toughness. 22. BlendelL. J. E. and Coble. R. L. Measurement of stress due to J.Am.Cerm.Soc.1992,75,3396-4000 expansion anisotropy in Al2 O3. J. Am. Cera. Soc., 198 12. Bueno, S, Moreno, R. and Baudin, C, Design and processing of 174-178 Al2O3-AlzTiOs layered structures. J. Eur: Ceram. Soc., 2005, 25, 23. De Portu, G. Micele. L and Pezzotti, G, Characterization of delam- ination cracks in Al2O3 /Al203+3Y-TZP multilayered composites by 13. Bueno. S, Moreno. R. and Baudin Reaction sintered Raman and fluorescence piezo-spectroscopy. Appl. Spectr, 2005, 59, Al]O3/Al TiOs microcrack-free co ation. J. Eur Ceram. Soc. 2004 5-279 24. Timoshenko, S. P, Mechanics of Materials. Van Nostrand Reinhold 14. Grabner, L, Spectroscopic techniq ne measurement of residual Co. New York, USA, 1972 stress in sintered Al2O3. J. Appl. 978,49,580-583
G. de Portu et al. / Journal of the European Ceramic Society 26 (2006) 2699–2705 2705 3. Bartolome, J., Requena, J., Moya, J. S., Li, M. and Guiu, F., Cyclic ´ fatigue crack growth resistance of Al2O3–Al2TiO5 composites. Acta Mater., 1996, 44, 1361–1370. 4. Bartolome, J., Requena, J., Moya, J. S., Li, M. and Guiu, F., Cyclic ´ fatigue of Al2O3–Al2TiO5 composites in direct push-pull. Fatigue Fract. Eng. Mater. Struct., 1997, 20, 789–798. 5. Uribe, R. and Baud´ın, C., Influence of a dispersion of aluminum titanate particles of controlled size on the thermal shock resistance of alumina. J. Am. Ceram. Soc., 2003, 86, 846–850. 6. Lawn, B. R., Padture, N. P., Braun, L. M. and Bennison, S. J., Model for toughness curves in two-phase ceramics: I, basic fracture mechanics. J. Am. Ceram. Soc., 1993, 76, 2235–2240. 7. Padture, N. P., Runyan, J. L., Bennison, S. J., Braun, L. M. and Lawn, B. R., Model for toughness curves in two-phase ceramics: II, microstructural variables. J. Am. Ceram. Soc., 1993, 76, 2241–2247. 8. Taylor, D., Thermal expansion data: III. Sesquioxides, M2O3 with the corundum and the A-, B- and C-M2O3. Br. Ceram. Trans. J., 1984, 83, 92–98. 9. Taylor, D., Thermal expansion data. XI. Complex oxides, A2BO5, and the garnets. Br. Ceram. Trans. J., 1987, 86, 1–6. 10. Chan, H. M., Layered ceramics: processing and mechanical behavior. Annu. Rev. Mater. Sci., 1997, 27, 249–282. 11. Russo, C. J., Harmer, M. P., Chan, H. M. and Miller, G. A., Design of a laminated ceramic composite for improved strength and toughness. J. Am. Ceram. Soc., 1992, 75, 3396–4000. 12. Bueno, S., Moreno, R. and Baud´ın, C., Design and processing of Al2O3–Al2TiO5 layered structures. J. Eur. Ceram. Soc., 2005, 25, 847–856. 13. Bueno, S., Moreno, R. and Baud´ın, C., Reaction sintered Al2O3/Al2TiO5 microcrack-free composites obtained by colloidal filtration. J. Eur. Ceram. Soc., 2004, 24, 2785–2791. 14. Grabner, L., Spectroscopic technique for the measurement of residual stress in sintered Al2O3. J. Appl. Phys., 1978, 49, 580–583. 15. De Portu, G., Micele, L., Sekiguchi, Y. and Pezzotti, G., Measurement of residual stress distributions in Al2O3/3Y-TZP multilayered composites by fluorescence and Raman microprobe piezo-spectroscopy. Acta Mater., 2005, 53, 1511–1520. 16. Ma, Q. and Clarke, D. R., Stress measurement in single-crystal and polycrystalline ceramics using their optical fluorescence. J. Am. Ceram. Soc., 1993, 76, 1433–1440. 17. Sergo, V., Lipkin, D. M., De Portu, G. and Clarke, D. R., Edge stresses in alumina/zirconia laminates. J. Am. Ceram. Soc., 1997, 80, 1633–1638. 18. Bueno, S. and Baud´ın, C., In situ developed laminates with large microstructural differences between layers of similar composition. J. Mater. Sci., in press. 19. Oel, H. J. and Frechette, V. D., Stress Distribution in multiphase ´ systems: I, composites with planar interfaces. J. Am. Ceram. Soc., 1967, 50, 542–549. 20. Ho, S., Hillman, C., Lange, F. F. and Suo, Z., Surface cracking in layers under biaxial, residual compressive stress. J. Am. Ceram. Soc., 1995, 78, 2353–2359. 21. Hasselman, D. P. H., Donaldson, K. Y., Anderson, E. M. and Johnson, T. A., Effect of thermal history on the thermal diffusivity and thermal expansion of an alumina-aluminium titanate composite. J. Am. Ceram. Soc., 1993, 76. 22. Blendell, J. E. and Coble, R. L., Measurement of stress due to thermal expansion anisotropy in Al2O3. J. Am. Ceram. Soc., 1982, 65(3), 174–178. 23. De Portu, G., Micele, L. and Pezzotti, G., Characterization of delamination cracks in Al2O3/Al2O3+3Y-TZP multilayered composites by Raman and fluorescence piezo-spectroscopy. Appl. Spectr., 2005, 59, 21–25. 24. Timoshenko, S. P., Mechanics of Materials. VanNostrand Reinhold Co, New York, USA, 1972