- Email: [email protected]
Characterization of rapid thermal processing oxynitrides by SIMS and XPS analyses
Applied Surface Science 144–145 Ž1999. 301–305
Characterization of rapid thermal processing oxynitrides by SIMS and XPS analyses M. Bersani ) , L. Vanzetti, M. Sbetti, M. Anderle Physics and Chemistry of Surface and Interface DiÕision, ITC-irst, 38050 PoÕo-Trento, Italy
Abstract Aim of this work is the characterisation of oxynitride films grown by rapid thermal processing ŽRTP. using nitrous oxide and nitric oxide precursors. The thickness of the oxynitrided layers is about 7 nm. Secondary ion mass spectrometry ŽSIMS. and X-ray photoemission spectroscopy ŽXPS. have been employed to obtain a complete chemical characterisation. XPS analyses have been performed at different depths after removal of oxynitride layers by chemical etching. SIMS and XPS analyses have been performed on the same samples after a reoxidation treatment as well. Depending on the precursors used, the oxynitrides show different characteristics. q 1999 Elsevier Science B.V. All rights reserved. PACS: 61.72.S; 81.70.J; 82.80.M; 79.60 Keywords: SIMS; XPS; Depth profiling; Nitrogen; Silicon oxide
1. Introduction The scaling down of integrated devices to submicron ranges has required new material process technologies. High-quality and ultrathin dielectrics for ULSI microelectronics technology are an example. Silicon oxynitride is known to have many advantages with respect to the conventional silicon dioxide w1x: high breakdown field, small charge trap and interface density of states increase, and boron penetration reduction. Nowadays, the oxynitrides are the
)
Corresponding author. Tel.: q39-0461-314485; Fax: q390461-810851; E-mail: [email protected]
preferred materials to obtain very thin dielectric films and are widely used for ULSI device manufacturing w2x. In this work, we perform an oxynitride characterisation using two complementary analytical methods: secondary ion mass spectrometry ŽSIMS. and X-ray photoemission spectroscopy ŽXPS.. SIMS analysis is commonly used in oxynitrided film depth profiling w3–5x; however, quantitative nitrogen distribution with subnanometer depth resolution is still an analytical challenge. XPS is employed to obtain both nitrogen profile and chemical state w5–7x, oxide nonuniformity and sensitivity being the major limitations. In addition, reoxidised samples have been analysed in order to evaluate the influence of the two
0169-4332r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 Ž 9 8 . 0 0 8 1 4 - 9
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
302
Sampled areas are 1 = 0.2 mm in size, and an overall energy resolution of 0.4 eV is routinely used. All the measurements were performed at a fixed emission angle of 908. Oxide films have been thinned down to obtain information on the whole nitrogen profile. Samples 2, 3, 4, and 5 were etched in a 1% HF solution, for times ranging from 0 to 90 s, and consequently the residual thickness ranged from 7 to 0.1 nm, as calculated from XPS data. The etching procedure was chosen versus Arq ion-gun sputtering to preserve chemical information from the samples. We have seen that Ar ions with energy as low as 1.5 eV, can induce surface artefacts. For each sample, Si 2p, O 1s, N 1s, C 1s, and F 1s core levels were collected. The carbon is due to surface contamination, and is not affecting the measurement of the buried interface. All core level peak energies were referenced to the elemental Si 2p 3r2 peak at 99.5 eV. The samples were all reoxidised for the same time and temperature conditions in order to evaluate the correlation between nitrogen content and thickness increase after reoxidation, and in particular to compare the equally nitrided NO and N2 O samples. The reoxidized samples have been analysed by SIMS and, after etching, by XPS.
different precursors on the reoxidation process and the consequent nitrogen redistribution.
2. Experimental Five oxynitride thin films have been prepared in a vertical furnace with a rapid thermal processing ŽRTP. treatment at the same temperature, four of them with NO precursor and different times of oxynitridation, and the last one with N2 O precursor ŽTable 1.. The film thickness was determined by ellipsometry measurements. SIMS analyses have been performed by Cameca 4f, bombarding with Csq primary beam, and monitoring MCsq secondary ions, where M is the species of interest Žin this case, N, O and Si.. The primary beam energy and sample bias have been fixed at 6.75 keV and 4.5 keV, respectively, yielding a primary beam impact energy of 2.25 keV. Under these conditions, the incidence angle is 608, and the depth resolution results to be less than 3 nm. Nitrogen profiles have been quantified using suitable standards: two N implants, in silicon and silicon dioxide matrix with 1 = 10 16 atomsrcm2 dose and 10 keV implant energy. The relative sensitivity factors ŽRSF. obtained in the analytical conditions described above, are 3 = 10 23 atomsrcm3 in silicon and 1 = 10 23 atomsrcm3 in silicon dioxide, respectively w4x. To quantify the nitrogen signal in the SiO 2rSi interface region, an RSF has been obtained using a function of the oxygen signal variation, as reported elsewhere w8,9x. XPS measurements have been carried out using a Scienta Esca-200 system equipped with a monochromatized Al K a Ž1486.6 eV. source.
3. Results and discussion Fig. 1a shows nitrogen SIMS profiles of samples 2, 4, and 5. Oxygen and silicon signals, also reported, prove the presence of three different zones into the film: SiO 2 , Si and an interface region pointed out by the slope variation in the silicon signal, not present in a pure silicon oxide film.
Table 1 Analyzed sample description Sample ID 1 2 3 4 5 a
Confidential data.
Precursor a
NO X percent NO X percent a NO X percent a NO 5 X percent a N2 O a
Time a
Thickness Žnm.
Total reoxidised thickness Žnm.
T1 2T1 4T1 4T1 1.5T1
7.09 6.92 7.03 6.96 7.05
29.7 28.7 26.9 23.0 16.0
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
303
Fig. 2. Nitrogen SIMS quantitative profiles on reoxidized samples 2 and 5.
Fig. 1. SIMS qualitative Ža. and quantitative Žb. profiles on samples 2, 4, and 5.
Total N integral, maximum concentration, position of the maximum and full width at half maximum ŽFWHM., are reported in Table 2. The FWHM values show a nitrided layer thickness below the SIMS depth resolution in all analysed samples. The quantitative nitrogen profiles, shown in Fig. 1b, are scaling down in peak concentration from samples 4 to 1, nitrided with NO precursor. Sample 5, nitrided with N2 O, shows an integral very close to sample 2, but a different shape. The regrown oxide thickness behaves as an inverse function of N concentration before the reoxida-
tion for the samples with NO precursor. Sample 5 does not fit this behaviour. In Fig. 2 nitrogen depth profiles of samples 2 and 5, taken after reoxidation, are reported. Both the peaks result to be at the same position, closest to the oxide surface with respect to the initial one, but the nitrogen peak concentrations after reoxidation are much lower than the initial ones. Despite the same initial nitrogen peak concentration, sample 5 shows a thinner regrown layer and an higher final nitrogen peak concentration with respect of sample 2. This indicates two different reoxidation kinetics. The atomic percentage of oxygen, nitrogen, and silicon for representative samples at two different residual oxide thicknesses is reported in Table 3. These values represent the integrated intensities of the XPS core level peaks, and are corrected for the atomic sensitivity factor of the various elements. The residual thickness after etching was calculated from XPS data, using the well established formula reported by Himpsel et al. w10x.
Table 2 SIMS results on samples before reoxidation Sample ID
N integral Žatomsrcm2 .
Maximum concentration Žatomsrcm3 .
FWHM Žnm.
1 2 3 4 5
7.0 = 10 14 1.2 = 10 15 1.5 = 10 15 2.0 = 10 15 1.1 = 10 15
2.9 = 0 21 4.5 = 10 21 6.3 = 10 21 7.5 = 10 21 3.3 = 10 21
2.5 2.4 2.0 2.7 2.8
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
304
Table 3 Oxygen, nitrogen, and silicon atomic concentration in samples 2, 3, 4, and 5, for two different oxide thicknesses: near the SiO 2 –Si interface, and in the ‘bulk’ Sample Oxygen Ž%. Nitrogen Ž%. Silicon Ž%. Thickness Žnm. 2 3 4 5 2 3 4 5
20.1 25.5 19.4 24.7 50.3 50.3 49.5 49.9
1.8 2.6 4.1 1.4 0.7 0.9 1.3 0.7
78.1 71.9 76.5 78.1 49.0 48.8 49.2 49.4
0.6 0.9 0.7 0.8 4.2 4.4 4.1 4
It is clear from Table 3 that the nitrogen concentration is higher near the interface, for all the samples, and that this concentration increases, from sample 2 to sample 4, and it is lower for sample 5, nitrided with a different precursor. To have a better idea of the behaviour of nitrogen in structures grown with different precursors, in Fig. 3 we show the nitrogen concentration depth profile for samples 4 and 5. These data, which have been calculated from the XPS peak intensities, show the wNx to wNx q wOx ratio in the oxide, as a function of distance from the Si–SiO 2 interface. The nitrogen distributions are clearly different for the two samples. Sample 4 has a peak in nitrogen concentration closer to the Si–SiO 2 interface, and this peak has a much higher value. These results confirm the reported SIMS results. In Fig. 4, we report the N 1s core level emission for samples 2, 3, 4, and 5, near the interface SiO 2 –Si.
Fig. 3. Nitrogen to oxygen plus nitrogen concentration ratio, as a function of residual oxide thickness for sample 4 Ždots. and sample 5 Žsquares..
Fig. 4. Nitrogen N 1s core emission for sample 2 Žsquare., 3 Žline., 4 Ždotted line., and 5 Žtriangle.. The binding energy scale is referenced to the Si 2p 3r 2 peak at 99.5 eV.
The core level lineshapes for the three samples grown with the NO precursor are clearly the same, and the peak height increases from sample 2 to sample 4, while the lineshape of the sample prepared with the N2 O precursor is quite different. Results for thicker oxides Žnot shown., are very similar. N 1s peaks from samples 2–4 have an asymmetric lineshape, with a full width at half maximum ŽFWHM. of 1.27 eV. They were deconvolved into two Gaussian peak, with consistent parameters. The first peak ŽN1. has a binding energy of 397.86 " 0.05 eV, and the second one ŽN2. a binding energy of 398.56 " 0.05 eV. The N1 binding energy is very close to that of N 1s in silicon nitride, therefore we can conclude that this peak is due to nitrogen atoms bound to three silicon atoms. The second peak has had many attributions in the literature w5,7,11,12x involving oxygen or hydrogen atoms, or both, but no definitive conclusion has been reached yet. N 1s peak from sample 5 has a larger FWHM Ž1.42 eV., and was deconvolved into three Gaussian peak, with consistent parameters. The first peak ŽN1. has a binding energy of 397.9 " 0.05 eV, again nitrogen atoms bound to three silicon atoms. The second ŽN2., and third ŽN3. peaks have binding energies of 398.50 " 0.05 eV, and 399.08 " 0.05 eV, respectively. As for the N2 peak in the NO samples, the attribution of these two peaks to specific bonds is still very controversial. Work is in progress to achieve a better understanding of the chemistry of this interface. Preliminary measurements were performed on reoxidized samples Ž2 and 5.. The XPS results Žnot
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
shown., indicate that the two samples have different oxide thickness Žas found in SIMS measurements., and that the nitrogen concentration is around or below the XPS detection limit. 4. Conclusions Oxynitrided samples grown by RTP with NO or N2 O precursors present differences in nitrogen amount, distribution and chemical state, as shown by XPS and SIMS analyses. Si 3 N4 has been clearly identified as nitrogen chemical compound. Precursor dependent reoxidation processes are also pointed out by these analytical methods. References w1x T. Hori, H. Hiroshi, K. Tsuji, IEEE Trans. Electron. Devices 36 Ž1989. 340.
305
w2x T. Arakawa, R. Matsumoto, T. Hayashi, J. Appl. Phys. 36 Ž1997. 1351. w3x Y. Okada, P.J. Tobin, V. Lakhotia, S.A. Ajuria, R.I. Hedge, J.C. Liao, P.P. Rushbrook, L.J. Arias Jr., J. Electrochem. Soc. 140 Ž1993. L87. w4x M.R. Frost, C.W. Magee, Appl. Surf. Sci. 104–105 Ž1996. 379. w5x D. Bouvet, P.A. Clivaz, M. Dutoit, C. Coluzza, J. Almeida, G. Margaritondo, F. Pio, J. Appl. Phys. 79 Ž1996. 7114. w6x A. Kamath, D.L. Kwong, Y.M. Sun, P.M. Blass, S. Whaley, J.M. White, J. Appl. Phys. Lett. 70 Ž1997. 63. w7x E.C. Carr, R.A. Buhrman, Appl. Phys. Lett. 63 Ž1993. 54. w8x M. Bersani, M. Fedrizzi, M. Ferroni, C. Savoia, M. Anderle, Secondary Ion Mass Spectrometry SIMS XI, Wiley, New York, 1998, p. 1055. w9x M. Bersani, M. Fedrizzi, M. Sbetti, M. Anderle, Characterization and Metrology for ULSI Technology, 1998, International Conference, Woodbury, NY, 1998, pp. 892–896. w10x F.J. Himpsel, F.R. McFeely, A. Taleb-Ibrahimi, J.A. Yarmoff, G. Hollinger, Phys. Rev. B 38 Ž1988. 6084. w11x Z.H. Lu, S.P. Tai, R. Kao, P. Pianetta, Appl. Phys. Lett. 67 Ž1995. 2836. w12x M. Bhat, G.W. Yoon, J. Kim, D.L. Kwong, M. Arendt, J.M. White, Appl. Phys. Lett. 64 Ž1994. 2116.
Characterization of rapid thermal processing oxynitrides by SIMS and XPS analyses M. Bersani ) , L. Vanzetti, M. Sbetti, M. Anderle Physics and Chemistry of Surface and Interface DiÕision, ITC-irst, 38050 PoÕo-Trento, Italy
Abstract Aim of this work is the characterisation of oxynitride films grown by rapid thermal processing ŽRTP. using nitrous oxide and nitric oxide precursors. The thickness of the oxynitrided layers is about 7 nm. Secondary ion mass spectrometry ŽSIMS. and X-ray photoemission spectroscopy ŽXPS. have been employed to obtain a complete chemical characterisation. XPS analyses have been performed at different depths after removal of oxynitride layers by chemical etching. SIMS and XPS analyses have been performed on the same samples after a reoxidation treatment as well. Depending on the precursors used, the oxynitrides show different characteristics. q 1999 Elsevier Science B.V. All rights reserved. PACS: 61.72.S; 81.70.J; 82.80.M; 79.60 Keywords: SIMS; XPS; Depth profiling; Nitrogen; Silicon oxide
1. Introduction The scaling down of integrated devices to submicron ranges has required new material process technologies. High-quality and ultrathin dielectrics for ULSI microelectronics technology are an example. Silicon oxynitride is known to have many advantages with respect to the conventional silicon dioxide w1x: high breakdown field, small charge trap and interface density of states increase, and boron penetration reduction. Nowadays, the oxynitrides are the
)
Corresponding author. Tel.: q39-0461-314485; Fax: q390461-810851; E-mail: [email protected]
preferred materials to obtain very thin dielectric films and are widely used for ULSI device manufacturing w2x. In this work, we perform an oxynitride characterisation using two complementary analytical methods: secondary ion mass spectrometry ŽSIMS. and X-ray photoemission spectroscopy ŽXPS.. SIMS analysis is commonly used in oxynitrided film depth profiling w3–5x; however, quantitative nitrogen distribution with subnanometer depth resolution is still an analytical challenge. XPS is employed to obtain both nitrogen profile and chemical state w5–7x, oxide nonuniformity and sensitivity being the major limitations. In addition, reoxidised samples have been analysed in order to evaluate the influence of the two
0169-4332r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 Ž 9 8 . 0 0 8 1 4 - 9
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
302
Sampled areas are 1 = 0.2 mm in size, and an overall energy resolution of 0.4 eV is routinely used. All the measurements were performed at a fixed emission angle of 908. Oxide films have been thinned down to obtain information on the whole nitrogen profile. Samples 2, 3, 4, and 5 were etched in a 1% HF solution, for times ranging from 0 to 90 s, and consequently the residual thickness ranged from 7 to 0.1 nm, as calculated from XPS data. The etching procedure was chosen versus Arq ion-gun sputtering to preserve chemical information from the samples. We have seen that Ar ions with energy as low as 1.5 eV, can induce surface artefacts. For each sample, Si 2p, O 1s, N 1s, C 1s, and F 1s core levels were collected. The carbon is due to surface contamination, and is not affecting the measurement of the buried interface. All core level peak energies were referenced to the elemental Si 2p 3r2 peak at 99.5 eV. The samples were all reoxidised for the same time and temperature conditions in order to evaluate the correlation between nitrogen content and thickness increase after reoxidation, and in particular to compare the equally nitrided NO and N2 O samples. The reoxidized samples have been analysed by SIMS and, after etching, by XPS.
different precursors on the reoxidation process and the consequent nitrogen redistribution.
2. Experimental Five oxynitride thin films have been prepared in a vertical furnace with a rapid thermal processing ŽRTP. treatment at the same temperature, four of them with NO precursor and different times of oxynitridation, and the last one with N2 O precursor ŽTable 1.. The film thickness was determined by ellipsometry measurements. SIMS analyses have been performed by Cameca 4f, bombarding with Csq primary beam, and monitoring MCsq secondary ions, where M is the species of interest Žin this case, N, O and Si.. The primary beam energy and sample bias have been fixed at 6.75 keV and 4.5 keV, respectively, yielding a primary beam impact energy of 2.25 keV. Under these conditions, the incidence angle is 608, and the depth resolution results to be less than 3 nm. Nitrogen profiles have been quantified using suitable standards: two N implants, in silicon and silicon dioxide matrix with 1 = 10 16 atomsrcm2 dose and 10 keV implant energy. The relative sensitivity factors ŽRSF. obtained in the analytical conditions described above, are 3 = 10 23 atomsrcm3 in silicon and 1 = 10 23 atomsrcm3 in silicon dioxide, respectively w4x. To quantify the nitrogen signal in the SiO 2rSi interface region, an RSF has been obtained using a function of the oxygen signal variation, as reported elsewhere w8,9x. XPS measurements have been carried out using a Scienta Esca-200 system equipped with a monochromatized Al K a Ž1486.6 eV. source.
3. Results and discussion Fig. 1a shows nitrogen SIMS profiles of samples 2, 4, and 5. Oxygen and silicon signals, also reported, prove the presence of three different zones into the film: SiO 2 , Si and an interface region pointed out by the slope variation in the silicon signal, not present in a pure silicon oxide film.
Table 1 Analyzed sample description Sample ID 1 2 3 4 5 a
Confidential data.
Precursor a
NO X percent NO X percent a NO X percent a NO 5 X percent a N2 O a
Time a
Thickness Žnm.
Total reoxidised thickness Žnm.
T1 2T1 4T1 4T1 1.5T1
7.09 6.92 7.03 6.96 7.05
29.7 28.7 26.9 23.0 16.0
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
303
Fig. 2. Nitrogen SIMS quantitative profiles on reoxidized samples 2 and 5.
Fig. 1. SIMS qualitative Ža. and quantitative Žb. profiles on samples 2, 4, and 5.
Total N integral, maximum concentration, position of the maximum and full width at half maximum ŽFWHM., are reported in Table 2. The FWHM values show a nitrided layer thickness below the SIMS depth resolution in all analysed samples. The quantitative nitrogen profiles, shown in Fig. 1b, are scaling down in peak concentration from samples 4 to 1, nitrided with NO precursor. Sample 5, nitrided with N2 O, shows an integral very close to sample 2, but a different shape. The regrown oxide thickness behaves as an inverse function of N concentration before the reoxida-
tion for the samples with NO precursor. Sample 5 does not fit this behaviour. In Fig. 2 nitrogen depth profiles of samples 2 and 5, taken after reoxidation, are reported. Both the peaks result to be at the same position, closest to the oxide surface with respect to the initial one, but the nitrogen peak concentrations after reoxidation are much lower than the initial ones. Despite the same initial nitrogen peak concentration, sample 5 shows a thinner regrown layer and an higher final nitrogen peak concentration with respect of sample 2. This indicates two different reoxidation kinetics. The atomic percentage of oxygen, nitrogen, and silicon for representative samples at two different residual oxide thicknesses is reported in Table 3. These values represent the integrated intensities of the XPS core level peaks, and are corrected for the atomic sensitivity factor of the various elements. The residual thickness after etching was calculated from XPS data, using the well established formula reported by Himpsel et al. w10x.
Table 2 SIMS results on samples before reoxidation Sample ID
N integral Žatomsrcm2 .
Maximum concentration Žatomsrcm3 .
FWHM Žnm.
1 2 3 4 5
7.0 = 10 14 1.2 = 10 15 1.5 = 10 15 2.0 = 10 15 1.1 = 10 15
2.9 = 0 21 4.5 = 10 21 6.3 = 10 21 7.5 = 10 21 3.3 = 10 21
2.5 2.4 2.0 2.7 2.8
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
304
Table 3 Oxygen, nitrogen, and silicon atomic concentration in samples 2, 3, 4, and 5, for two different oxide thicknesses: near the SiO 2 –Si interface, and in the ‘bulk’ Sample Oxygen Ž%. Nitrogen Ž%. Silicon Ž%. Thickness Žnm. 2 3 4 5 2 3 4 5
20.1 25.5 19.4 24.7 50.3 50.3 49.5 49.9
1.8 2.6 4.1 1.4 0.7 0.9 1.3 0.7
78.1 71.9 76.5 78.1 49.0 48.8 49.2 49.4
0.6 0.9 0.7 0.8 4.2 4.4 4.1 4
It is clear from Table 3 that the nitrogen concentration is higher near the interface, for all the samples, and that this concentration increases, from sample 2 to sample 4, and it is lower for sample 5, nitrided with a different precursor. To have a better idea of the behaviour of nitrogen in structures grown with different precursors, in Fig. 3 we show the nitrogen concentration depth profile for samples 4 and 5. These data, which have been calculated from the XPS peak intensities, show the wNx to wNx q wOx ratio in the oxide, as a function of distance from the Si–SiO 2 interface. The nitrogen distributions are clearly different for the two samples. Sample 4 has a peak in nitrogen concentration closer to the Si–SiO 2 interface, and this peak has a much higher value. These results confirm the reported SIMS results. In Fig. 4, we report the N 1s core level emission for samples 2, 3, 4, and 5, near the interface SiO 2 –Si.
Fig. 3. Nitrogen to oxygen plus nitrogen concentration ratio, as a function of residual oxide thickness for sample 4 Ždots. and sample 5 Žsquares..
Fig. 4. Nitrogen N 1s core emission for sample 2 Žsquare., 3 Žline., 4 Ždotted line., and 5 Žtriangle.. The binding energy scale is referenced to the Si 2p 3r 2 peak at 99.5 eV.
The core level lineshapes for the three samples grown with the NO precursor are clearly the same, and the peak height increases from sample 2 to sample 4, while the lineshape of the sample prepared with the N2 O precursor is quite different. Results for thicker oxides Žnot shown., are very similar. N 1s peaks from samples 2–4 have an asymmetric lineshape, with a full width at half maximum ŽFWHM. of 1.27 eV. They were deconvolved into two Gaussian peak, with consistent parameters. The first peak ŽN1. has a binding energy of 397.86 " 0.05 eV, and the second one ŽN2. a binding energy of 398.56 " 0.05 eV. The N1 binding energy is very close to that of N 1s in silicon nitride, therefore we can conclude that this peak is due to nitrogen atoms bound to three silicon atoms. The second peak has had many attributions in the literature w5,7,11,12x involving oxygen or hydrogen atoms, or both, but no definitive conclusion has been reached yet. N 1s peak from sample 5 has a larger FWHM Ž1.42 eV., and was deconvolved into three Gaussian peak, with consistent parameters. The first peak ŽN1. has a binding energy of 397.9 " 0.05 eV, again nitrogen atoms bound to three silicon atoms. The second ŽN2., and third ŽN3. peaks have binding energies of 398.50 " 0.05 eV, and 399.08 " 0.05 eV, respectively. As for the N2 peak in the NO samples, the attribution of these two peaks to specific bonds is still very controversial. Work is in progress to achieve a better understanding of the chemistry of this interface. Preliminary measurements were performed on reoxidized samples Ž2 and 5.. The XPS results Žnot
M. Bersani et al.r Applied Surface Science 144–145 (1999) 301–305
shown., indicate that the two samples have different oxide thickness Žas found in SIMS measurements., and that the nitrogen concentration is around or below the XPS detection limit. 4. Conclusions Oxynitrided samples grown by RTP with NO or N2 O precursors present differences in nitrogen amount, distribution and chemical state, as shown by XPS and SIMS analyses. Si 3 N4 has been clearly identified as nitrogen chemical compound. Precursor dependent reoxidation processes are also pointed out by these analytical methods. References w1x T. Hori, H. Hiroshi, K. Tsuji, IEEE Trans. Electron. Devices 36 Ž1989. 340.
305
w2x T. Arakawa, R. Matsumoto, T. Hayashi, J. Appl. Phys. 36 Ž1997. 1351. w3x Y. Okada, P.J. Tobin, V. Lakhotia, S.A. Ajuria, R.I. Hedge, J.C. Liao, P.P. Rushbrook, L.J. Arias Jr., J. Electrochem. Soc. 140 Ž1993. L87. w4x M.R. Frost, C.W. Magee, Appl. Surf. Sci. 104–105 Ž1996. 379. w5x D. Bouvet, P.A. Clivaz, M. Dutoit, C. Coluzza, J. Almeida, G. Margaritondo, F. Pio, J. Appl. Phys. 79 Ž1996. 7114. w6x A. Kamath, D.L. Kwong, Y.M. Sun, P.M. Blass, S. Whaley, J.M. White, J. Appl. Phys. Lett. 70 Ž1997. 63. w7x E.C. Carr, R.A. Buhrman, Appl. Phys. Lett. 63 Ž1993. 54. w8x M. Bersani, M. Fedrizzi, M. Ferroni, C. Savoia, M. Anderle, Secondary Ion Mass Spectrometry SIMS XI, Wiley, New York, 1998, p. 1055. w9x M. Bersani, M. Fedrizzi, M. Sbetti, M. Anderle, Characterization and Metrology for ULSI Technology, 1998, International Conference, Woodbury, NY, 1998, pp. 892–896. w10x F.J. Himpsel, F.R. McFeely, A. Taleb-Ibrahimi, J.A. Yarmoff, G. Hollinger, Phys. Rev. B 38 Ž1988. 6084. w11x Z.H. Lu, S.P. Tai, R. Kao, P. Pianetta, Appl. Phys. Lett. 67 Ž1995. 2836. w12x M. Bhat, G.W. Yoon, J. Kim, D.L. Kwong, M. Arendt, J.M. White, Appl. Phys. Lett. 64 Ž1994. 2116.