G. de Portu et al. Acta Materialia 53(2005)1511-1520 Cross-sections of the laminates were ground and pol- where lli is the trace of the piezo-spectroscopic matrix ished with successively finer diamond pastes until opti- and oi is the first invariant of the stress tensor(aj/ lly flat surfaces were produced, suitable for being commonly referred to as the mean normal stress) spectroscopIc assessments Therefore, if Ii is known, the spectral shift Av can be As reference materials. a series of eight 3Y-TZP. considered to be a direct measure of the normal stres containing ceramics were investigated: a single-phase within the volume probed by the laser beam for each 3Y-TZP, and seven bi-phase Al2O3/ZrO2 materials, con- spectra acquisition. In laminate structures, far away taining 3, 10, 20, 50, 70, 90, 97 wt% Al2O3. For compar- from external edges [6], the residual stress field can be on,a dense monolithic alumina sample was also considered to be of a biaxial nature, accordingly, it prepared. The Al2 O3/ZrOz samples were sintered at should be computed as 3ay On the other hand, nearby 1550C for I h. Samples were ground and polished the edges the residual stress field is typically three- according to the same procedure adopted for the dimensional, thus requiring in Eq (1) the use of a mean cross-sections of multilayered specimens normal stress o 2.2. Applied and residual stress measurements 3. Results and discussion Microscopic stress distributions were measured by 3. 1. Piezo-spectroscopic calibrations and their reliability fluorescence and Raman piezo-spectroscopic methods by collecting either linear or two-dimensional arrays of spectra on the specimen side or cross-sections. The auto- Figs. 1(a)and(b)show typical fluorescence and Ra matically collected arrays of spectra in relatively large man spectra of Al2O3 and tetragonal ZrO2, respectively two-dimensional maps were typically 2 um-spaced and The stress dependences of the R, fluorescence band of the laser spot size was about 5 um (.e, using a 20x opti- chromophoric Al2O3 and of the 460 cm Raman band cal lens). Linear maps were collected either in the same of monolithic ZrO, as recorded in unconstrained mono- configuration of the two-dimensional maps or with lithic bodies are also shown in Fig. I(c)and( d), respec- I Am spacing and laser spot size of I um. Specimens tively. From a comparison between Figs. 1(c)and(d), were placed on a mapping device(lateral resolution of two features can be envisaged: (i) the trace piezo-spec 0.01 um), which was connected to a personal computer troscopic tensor, Ii, of the 460 cm ZrO2 band is to drive highly precise displacements(along both X about a half that of the R, fluorescence band of chromo and Y axes)on the specimen surface. The optical micro- phonic AlO3; (ii)a higher data scatter is envisaged for scope was used to both stimulate and collect the excited the trace piezo-spectroscopic tensor, i, of the fluorescence or Raman bands, which were analyzed 460 cm-I ZrO, band. Both circumstances may involve using an attached spectrometer (T-64000, Horiba/Jo- significant error in stress evaluation by the 460 cm bin-Ivon). An Ar-ion laser operating at a wavelength ZrO2 Raman band as compared to the Ri band of of 488 nm was used as the excitation source. In each Al2O3. The scatter of the data plot in Fig. 1(d), which map, the region of interest was first selected using the may arise from both inhomogeneity of the Y2O3 dopant optical microscope. Spectra were collected using a distribution and lack of intensity/sharpness of the ZrO2 CCD camera with an integration time of l and 4 s for Raman bands(as compared to the fluorescence bands of Al2O3 and ZrO2, respectively. Bands from a Hg/Ne Al,O3), reflects a confidence in the stress measurement lamp were used as an internal reference for spectral cal- of the order +100 MPa for the zirconia band, against ibrations. The collected data were automatically ana- the scatter +10 MPa allowed by the R fluorescence ed with curve fitting algorithms included in the band of alumina(Fig. I(c). Based on these preliminary LabSpec software package(LABSPEC 4.02, Horiba a/ assessments. we decided to develo perimental pr Jobin-lvon). Calibrations of spectral shift vs. externally cedure which, based only on the Ri fluorescence line of applied stress were made using a miniature four-point Al,O3, could enable us to extract the magnitude of the bending jig connected with a load-cell to measure in situ macroscopic stress field stored within layers avoiding the applied load. Details of the calibration procedures the large error involved in measuring the spectral shift were given in a previous report [5]. In a polycrystalline of Y-TZP Raman bands. Fig. 2(a) shows the shift(as ample(having no significant texture and a fine grained averaged on about 1000 measurements)measured for microstructure), the spectral shift, Av, of the Cr fluo- the R band of Al2O3 when different volume fractions rescence lines in AlO3(RI and R2 lines, henceforth) of Al2O phase were incorporated into a 3Y-TZP and of the Raman bands of 3Y-TzP under uniaxial tered body. In this preliminary spectroscopic assess stress can be expressed, to a linear approximation, as: ment, the specimens were neither assembled in layered △v=n structures nor subjected to external load, therefore the measured shifts merely represented the effect ofCross-sections of the laminates were ground and pol￾ished with successively finer diamond pastes until opti￾cally flat surfaces were produced, suitable for spectroscopic assessments. As reference materials, a series of eight 3Y-TZP￾containing ceramics were investigated: a single-phase 3Y-TZP, and seven bi-phase Al2O3/ZrO2 materials, con￾taining 3, 10, 20, 50, 70, 90, 97 wt% Al2O3. For compar￾ison, a dense monolithic alumina sample was also prepared. The Al2O3/ZrO2 samples were sintered at 1550 C for 1 h. Samples were ground and polished according to the same procedure adopted for the cross-sections of multilayered specimens. 2.2. Applied and residual stress measurements Microscopic stress distributions were measured by fluorescence and Raman piezo-spectroscopic methods by collecting either linear or two-dimensional arrays of spectra on the specimen side or cross-sections. The auto￾matically collected arrays of spectra in relatively large two-dimensional maps were typically 2 lm-spaced and the laser spot size was about 5 lm (i.e., using a 20· opti￾cal lens). Linear maps were collected either in the same configuration of the two-dimensional maps or with 1 lm spacing and laser spot size of 1 lm. Specimens were placed on a mapping device (lateral resolution of 0.01 lm), which was connected to a personal computer to drive highly precise displacements (along both X and Y axes) on the specimen surface. The optical micro￾scope was used to both stimulate and collect the excited fluorescence or Raman bands, which were analyzed using an attached spectrometer (T-64000, Horiba/Jo￾bin-Ivon). An Ar-ion laser operating at a wavelength of 488 nm was used as the excitation source. In each map, the region of interest was first selected using the optical microscope. Spectra were collected using a CCD camera with an integration time of 1 and 4 s for Al2O3 and ZrO2, respectively. Bands from a Hg/Ne lamp were used as an internal reference for spectral cal￾ibrations. The collected data were automatically ana￾lyzed with curve fitting algorithms included in the LabSpec software package (LABSPEC 4.02’’, Horiba/ Jobin-Ivon). Calibrations of spectral shift vs. externally applied stress were made using a miniature four-point bending jig connected with a load-cell to measure in situ the applied load. Details of the calibration procedures were given in a previous report [5]. In a polycrystalline sample (having no significant texture and a fine grained microstructure), the spectral shift, Dm, of the Cr3+ fluo￾rescence lines in Al2O3 (R1 and R2 lines, henceforth) and of the Raman bands of 3Y-TZP under uniaxial stress can be expressed, to a linear approximation, as: Dm ¼ 1 3 Piirjj; ð1Þ where Pii is the trace of the piezo-spectroscopic matrix and rjj is the first invariant of the stress tensor (rjj/3 being commonly referred to as the mean normal stress). Therefore, if Pii is known, the spectral shift Dm can be considered to be a direct measure of the normal stress within the volume probed by the laser beam for each spectra acquisition. In laminate structures, far away from external edges [6], the residual stress field can be considered to be of a biaxial nature; accordingly, it should be computed as 2 3 rjj. On the other hand, nearby the edges the residual stress field is typically three￾dimensional, thus requiring in Eq. (1) the use of a mean normal stress rjj. 3. Results and discussion 3.1. Piezo-spectroscopic calibrations and their reliability assessment Figs. 1(a) and (b) show typical fluorescence and Ra￾man spectra of Al2O3 and tetragonal ZrO2, respectively. The stress dependences of the R1 fluorescence band of chromophoric Al2O3 and of the 460 cm
1 Raman band of monolithic ZrO2 as recorded in unconstrained mono￾lithic bodies are also shown in Fig. 1(c) and (d), respec￾tively. From a comparison between Figs. 1(c) and (d), two features can be envisaged: (i) the trace piezo-spec￾troscopic tensor, Pii, of the 460 cm
1 ZrO2 band is about a half that of the R1 fluorescence band of chromo￾phoric Al2O3; (ii) a higher data scatter is envisaged for the trace piezo-spectroscopic tensor, Pii, of the 460 cm
1 ZrO2 band. Both circumstances may involve significant error in stress evaluation by the 460 cm
1 ZrO2 Raman band as compared to the R1 band of Al2O3. The scatter of the data plot in Fig. 1(d), which may arise from both inhomogeneity of the Y2O3 dopant distribution and lack of intensity/sharpness of the ZrO2 Raman bands (as compared to the fluorescence bands of Al2O3), reflects a confidence in the stress measurement of the order ±100 MPa for the zirconia band, against the scatter ±10 MPa allowed by the R1 fluorescence band of alumina (Fig. 1(c)). Based on these preliminary assessments, we decided to develop an experimental pro￾cedure which, based only on the R1 fluorescence line of Al2O3, could enable us to extract the magnitude of the macroscopic stress field stored within layers avoiding the large error involved in measuring the spectral shift of Y-TZP Raman bands. Fig. 2(a) shows the shift (as averaged on about 1000 measurements) measured for the R1 band of Al2O3 when different volume fractions of Al2O3 phase were incorporated into a 3Y-TZP sin￾tered body. In this preliminary spectroscopic assess￾ment, the specimens were neither assembled in layered structures nor subjected to external load, therefore the measured shifts merely represented the effect of G. de Portu et al. / Acta Materialia 53 (2005) 1511–1520 1513