A B C D Figure 2 Fracture surface of laminate composite(A)Si3N4/Si3N4 laminates;(B)Si3 Na/Si3N4-20 wt %TiN laminates;(C) Si3 Na/2(Si3N4 20 wt %TIN) laminates and(D) Si3 N4/TiN laminates. during crack propagation has also been estimated [25, however. The spectrum in Fig 5b was taken from the 26]. Some attempts to use Raman spectroscopy to es- center of a Vickers indentation(20 kg load) placed in timate the residual stresses around indentation in sil- the center of a thin Si3N4 layer from the same face icon nitride have been done [27, 28], but the results of Si3 N4/2(Si3 N4-20%TiN) laminate. The first three were contradictive and further clarification is needed. bands remain intact, but the other bands shifted to the The determination of residual stress in laminates is a higher wave numbers, which indicates the existence omplicated problem. Here we report the preliminary of a residual compressive stress in the center of the results of the Raman shift measurements that can be Vickers impression induced by the indentation. These further used to estimate the residual stresses in a lami- results are similar to published results [27, 28 nar composite One-dimensional maps of band shift, band inten- Two typical Si3 N4 Raman spectra are shown in Fig ity,FWHM, and other band parameters can be pro The spectrum in Fig 5a was taken at the center of a thin duced using a line scan technique [29]. Line map- Si, N4 layer from the side face of the Si3 N4/2(Si3 N4-20 ping of the 862 cm-I Raman band of silicon ni- wt%TiN) laminate. The thickness of Si3N4 layer was tride was performed across a thin Si3N4 layer from about 250 um and the thickness of Si3N4-20 wt%tin the Si3N4/2(Si3N4-20%TiN) laminate, starting at the layer was about 500 um. A first indication of existing Si3N4-20%TiN layer, crossing the interfaces, and end tensile mean stress came from the shift of the Si band to ing in the next Si3N4-20%TiN layer(Fig. 6). Maps of 518cm- because the spectrometer was calibrated w intensity(Fig 6A), FWHM(Fig. 6B), and peak shift a Si band being at 520.3 cm- at the beginning of exper-(Fig. 6C) were generated. As one can see, there is a iment. Free Si can sometimes be detected in Si3 N4 as shift in peak position from 862. 54 cm-in the Si N4 a result of desublimation of Si3N4. The band positions 20%TiN layer to 861.05 cm-I in the pure Si3N4 layer of unstressed Si3N4 were determined as 181, 203, 224,(Fig. 6C). Similar results have been published in [30] 446, 615, 728, 862, 926, 936, 1044 cm-. Also, Si3N4 The shift exists because of different surface stress states bands 862 cm,1044 cm, and others are shifted in layers with different composition [31-33]. There to lower wavenumbers in the center of a thin Si3N4 is tensile mean stress on the surface of the si3N4 layer. Three strong bands(181, 203, 224 cm)did not layers, since a down shift of the peak position was change their positions relative to the unstressed Si3N4, found. At the same time, a compressive mean stressFigure 2 Fracture surface of laminate composite. (A) Si3N4/Si3N4 laminates; (B) Si3N4/Si3N4-20 wt.%TiN laminates; (C) Si3N4/2(Si3N4- 20 wt.%TiN) laminates and (D) Si3N4/TiN laminates. during crack propagation has also been estimated [25, 26]. Some attempts to use Raman spectroscopy to es￾timate the residual stresses around indentation in sil￾icon nitride have been done [27, 28], but the results were contradictive and further clarification is needed. The determination of residual stress in laminates is a complicated problem. Here we report the preliminary results of the Raman shift measurements that can be further used to estimate the residual stresses in a lami￾nar composite. Two typical Si3N4 Raman spectra are shown in Fig. 5. The spectrum in Fig. 5a was taken at the center of a thin Si3N4 layer from the side face of the Si3N4/2(Si3N4-20 wt%TiN) laminate. The thickness of Si3N4 layer was about 250 µm and the thickness of Si3N4-20 wt%TiN layer was about 500 µm. A first indication of existing tensile mean stress came from the shift of the Si band to 518 cm−1 because the spectrometer was calibrated with a Si band being at 520.3 cm−1 at the beginning of exper￾iment. Free Si can sometimes be detected in Si3N4 as a result of desublimation of Si3N4. The band positions of unstressed Si3N4 were determined as 181, 203, 224, 446, 615, 728, 862, 926, 936, 1044 cm−1. Also, Si3N4 bands 862 cm−1, 1044 cm−1, and others are shifted to lower wavenumbers in the center of a thin Si3N4 layer. Three strong bands (181, 203, 224 cm−1) did not change their positions relative to the unstressed Si3N4, however. The spectrum in Fig. 5b was taken from the center of a Vickers indentation (20 kg load) placed in the center of a thin Si3N4 layer from the same face of Si3N4/2(Si3N4-20%TiN) laminate. The first three bands remain intact, but the other bands shifted to the higher wave numbers, which indicates the existence of a residual compressive stress in the center of the Vickers impression induced by the indentation. These results are similar to published results [27, 28]. One-dimensional maps of band shift, band inten￾sity, FWHM, and other band parameters can be pro￾duced using a line scan technique [29]. Line map￾ping of the 862 cm−1 Raman band of silicon ni￾tride was performed across a thin Si3N4 layer from the Si3N4/2(Si3N4-20%TiN) laminate, starting at the Si3N4-20% TiN layer, crossing the interfaces, and end￾ing in the next Si3N4-20%TiN layer (Fig. 6). Maps of intensity (Fig. 6A), FWHM (Fig. 6B), and peak shift (Fig. 6C) were generated. As one can see, there is a shift in peak position from 862.54 cm−1 in the Si3N4- 20%TiN layer to 861.05 cm−1 in the pure Si3N4 layer (Fig. 6C). Similar results have been published in [30]. The shift exists because of different surface stress states in layers with different composition [31–33]. There is tensile mean stress on the surface of the Si3N4 layers, since a down shift of the peak position was found. At the same time, a compressive mean stress 5447