Analysis of Structural Distribution of Nitrogen-Incorporated Species at

Anal. Chem. 2008, 80, 6286–6292

Analysis of Structural Distribution of Nitrogen-Incorporated Species at the Interface of Silicon Oxide Films on Silicon Using Time-of-Flight Secondary Ion Mass Spectrometry and Poisson Approximation Kiyoshi Chiba* Department of Nano Material and Bio Engineering, Tokushima Bunri University, Sanuki, Kagawa 769-2193, Japan Structural distributions of nitrogen-incorporated species at the interface between silicon and SiO2 films for plasmaenhanced chemical vapor deposited (PECVD) SiO2 films have been analyzed using time-of-flight secondary ion mass spectrometry (TOF-SIMS) and the Poisson approximation. Depth profiles of the intensity of Si3+ cluster ions were used to determine the depth-resolution function of the analytical system that causes beam-induced broadening effects in the depth profiles. The full width at halfmaximum (fwhm) of the resolution function was approximately 1.6 nm for 22 kV Au3+ primary ions and sputter etching by Ar+ at 1 kV. The intensity profile of Si3O2N+ secondary ion species at the interface was analyzed by assuming a Poisson distribution with a series of δ-layers. Convolution of the postulated signals with the depth resolution function was carried out to simulate depth profiles of Si3O2N+ ion species, which exhibit slow leading and sharp trailing edges. The analysis showed that the thickness of the δ-layer associated with the most probable existence of nitrogen-incorporated species is approximately 0.8 nm, and the fragmented species decreased from the interface to the SiO2 bulk film with a decay length of approximately 1.5 nm. Time-of-flight secondary ion mass spectrometry (TOF-SIMS) is a useful tool to characterize solid structures by detecting secondary ion species emitted from solid materials, by using a pulsed mode of bombardment of primary ions.1 It has the potential to play a powerful role in providing information on a few monolayers on the surface and depth profiles on the atomic scale in combination with surface etching techniques.2,3 It, however, sacrifices atomic-level information due to a number of factors causing the original information to be obscured.4,5 Although secondary ion species can be emitted from one or two atomic layers of a material surface, information collected by detectors at * Contact information. E-mail: [email protected]. Fax: +81-87-894-4201. (1) Briggs, D., Seah, M. P., Eds. Practical Surface Analysis, Vol. 2, 2nd ed.; Wiley & Sons Ltd.: Sussex, U.K., 1992. (2) Chiba, K.; Akamatsu, T.; Kawamura, M. Chem. Phys. Lett. 2006, 419, 506– 510. (3) Chiba, K.; Takenaka, Y. Appl. Surf. Sci. 2008, 254, 2534–2539. (4) Schulz, F.; Wittmaack, K.; Maul, J. Radiat. Effects 1973, 18, 211–215. (5) Hofmann, S. Appl. Surf. Sci. 1993, 70/71, 9–19.

6286

Analytical Chemistry, Vol. 80, No. 16, August 15, 2008

each depth may be limited to altered and averaged data; in other words, they are averaged over several atomic layers due to analytical limitations caused by instrumental and ion beam-induced factors.6,7 The instrumental factors can most likely be attributed to the unevenness of the crater bottom produced by the spatially varying removal rate on surface atomic layers during etching processes.8 This influence can be decreased to less than 1 nm using raster scan techniques with low energy ion bombardment for sputter etching.9 The ion beam induced factors may arise from atomic mixing which changes atomic positions by the bombardment of impinging sputter ions and selective sputtering.7,9–11 These factors lead to blurring of information and broadening of the depth profiles, which become obstacles to understanding material structures on an atomic scale. Beam-induced broadening effects in sputter depth profiling have been studied previously by a number of researchers.4–14 Random cascade mixing by ion bombardment can relocate atoms as almost isotropic displacements; this leads to symmetrical mixing distribution from a delta function of an original profile in very thin films.9 Taking into account simple elementary recoil events, recoil intermixing,4,6,12 attributed to differences in the primary collision (forward) and two-atom recoil processes (backward), induces an asymmetrical distribution with a fast exponential leading slope and a slow trailing slope in the depth profile; these slopes are identified by characteristic decay lengths in the leading and trailing edges, respectively.11,14 Further, it is to be noted that decay profiles depend on elements due to selective sputtering by ion bombardment during sputter etching.11 Although this beaminduced broadening in the depth profiles can be decreased using a primary ion beam with low beam energy, for instance less than 1 kV,9 it is important to maintain a practical sputter etching speed (6) Williams, P.; Baker, J. E. Appl. Phys. Lett. 1980, 36, 842–845. (7) Wittmaack, K.; Wach, W. Nucl. Instrum. Methods 1981, 191, 327–334. (8) Werner, H. W.; Hornstra, J.; Warmoltz, N. Surf. Interface Anal. 1983, 5, 87–88. (9) Wittmaack, K. Vacuum 1984, 34, 119–137. (10) Williams, P.; Baker, J. Nucl. Instrum. Methods 1981, 182/183, 15–24. (11) Wittmaack, K. Nucl. Instrum. Methods Phys. Res., Sect. B 1985, 7/8, 750– 754. (12) Littmark, U.; Hofer, W. O. Nucl. Instrum. Methods 1980, 168, 329–342. (13) Shepherd, F. R.; Robinson, W. H.; Brown, J. D.; Phillips, B. F. J. Vac. Sci. Technol., A 1983, 1, 991–994. (14) Turner, J. E.; Amano, J.; Gronet, C. M.; Gibbons, J. F. Appl. Phys. Lett. 1987, 50, 1601–1603. 10.1021/ac800675j CCC: $40.75  2008 American Chemical Society Published on Web 07/22/2008

and analyze the depth profiles and structural information close to the original ones, preferably with atomic layer by atomic layer resolution. It is generally accepted that the observed depth profile of intensity of a signal I(x) is given mathematically by the convolution integral of the original depth profile I0(x) with the instrumental depth resolution functionG(x), that is, the response function of a system as

I(x) )

∫ I (x′)G(x - x′) dx′ 0

(1)

where x is the depth.1 The procedure of deconvolution of the signal becomes a key issue to solve the equation and determine the original profile.15 Deconvolution of SIMS depth profiles has been studied by a number of researchers:16–35 including analytical procedures18–21,23–29 and implementation to δ-doped layers of boron (B) in Si,18,22,26,27,30–32,34,35 antimony (Sb),18 germanium (Ge),18 arsenic (As) in Si,33 and Si in GaAs.17,24,33 Nevertheless, it is not easy to deconvolute an observed signal directly to the original signal mainly due to difficulties in the mathematical procedures associated with noise and variations in signals. Thus, both the appropriate determination of a depth resolution function and modeling of depth distributions may help to resolve real experimental issues. Silicon oxynitride films have attracted increased recent interest as ultrathin dielectric films on silicon for gate oxides in semiconductor devices.36 Accumulation of the incorporated nitrogen at the Si-SiO2 interface often occurs and has been reported by a number of researchers.36–41 For fabrication of metal-oxide(15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35) (36) (37)

Ho, P. S.; Lewis, J. E. Surf. Sci. 1976, 55, 335–348. Sanz, J. M. Surf. Interface Anal. 1984, 6, 196. Clegg, J. B.; Beall, R. B. Surf. Interface Anal. 1989, 14, 307–314. Dowsett, M. G.; Barlow, R. D.; Fox, H. S.; Kubiak, R. A. A.; Collins, R. J. Vac. Sci. Technol., B 1992, 10, 336–341. Makarov, V. V. Surf. Interface Anal. 1993, 20, 821–826. Zalm, P. C.; de Kruif, R. C. M. Appl. Surf. Sci. 1993, 70/71, 73–78. Dowsett, M. G.; Rowlands, G.; Allen, P. N.; Barlow, R. D. Surf. Interface Anal. 1994, 21, 310–315. Herzel, F.; Ehwald, K. E.; Heinemann, B.; Kru ¨ ger, D.; Kurps, R.; Ropke, W.; Zeindl, H. P. Surf. Interface Anal. 1995, 23, 764–770. Makarov, V. V. Surf. Interface Anal. 1995, 23, 899. Cooke, G. A.; Dowsett, M. G.; Allen, P. N.; Collins, R.; Miethe, K. J. Vac. Sci. Technol., B 1996, 14, 132–135. Cooke, G. A.; Dowsett, M. G.; Phillips, P. J. Vac. Sci. Technol., B 1996, 14, 283–286. Gautier, B.; Dupuy, J. C.; Prost, R.; Prudon, G. Surf. Interface Anal. 1997, 25, 464–477. Chu, D. P.; Dowsett, M. G. Phys. Rev. B 1997, 56, 15167–15170. Dowsett, M. G.; Chu, D. P. J. Vac. Sci. Technol., B 1998, 16, 377–381. Gautier, B.; Prudon, G.; Dupuy, J. C. Surf. Interface Anal. 1998, 26, 974– 983. Gautier, B.; Dupuy, J. C.; Semmache, B.; Prudon, G. Nucl. Instrum. Methods Phys. Res., Sect. B 1998, 142, 361–376. Lee, J. W.; Kim, K. J.; Kim, H. K.; Moon, D. W. J. Surf. Anal. 2003, 10, 16–18. Shao, L.; Liu, J.; Wang, C.; Ma, K. B.; Zhang, J.; Chen, J.; Tang, D.; Patel, S.; Chu, W. K. Appl. Phys. Lett. 2003, 83, 5467–5469. Lee, J. W.; Kim, K. J.; Kim, H. K.; Moon, D. W. Surf. Interface Anal. 2005, 37, 176–180. Yang, M. H.; Goodman, G. G. Thin Solid Films 2006, 508, 276–278. Fares, B.; Gautier, B.; Dupuy, J. C.; Prudon, G.; Holliger, P. Appl. Surf. Sci. 2006, 252, 6478–6481. Green, M. L.; Gusev, E. P.; Degraeve, R.; Garfunkel, E. L. J. Appl. Phys. 2001, 90, 2057–2121. Saks, N. S.; Simons, M.; Fleetwood, D. M.; Yount, J. T.; Lenahan, P. M.; Klein, R. B. IEEE Trans. Nucl. Sci. 1994, 41, 1854–1863.

semiconductor (MOS) devices, nitrogen is incorporated into SiO2 films using either thermal oxynitridation or annealing or by chemical or physical deposition methods.36,42–44 The characterization of the incorporated nitrogen in the SiO2 films at the nanoscale is key to obtaining further knowledge of the chemical and structural features of these films and to pursue tailored nitrogen profiles in ultra thin dielectrics.45 In this paper, the author reports an analysis of the structural distribution of nitrogen-incorporated species at the interface between silicon and SiO2 films for plasma-enhanced chemical vapor deposited (PECVD) SiO2 films using time-of-flight secondary ion mass spectrometry and deconvolution of depth profiles with the Poisson approximation model. Both Si3+ cluster ions and Si3O2N+ secondary ion species from the Si-SiO2 interface were investigated. EXPERIMENTAL SECTION PECVD-prepared SiO2 films on single-crystal n-type (100) silicon wafers (resistivity of 4-6 Ω cm) were used in this study. PECVD-prepared SiO2 films were grown from SiH4 and N2O mixtures at a flow rate ratio of 1:18 at 400 °C after precleaning of the wafer with dilute HF and a mixture solution of trimethyl-2hydroxyethylammonium hydroxide (choline) and hydrogen peroxide (H2O2) without any postannealing. TOF-SIMS measurements were performed with an ULVACPHI TRIFT III TOF-SIMS spectrometer. Sputter etching was accomplished by Ar+ ions at 1 kV and 200 nA rastered over a 1000 × 1000 µm2 area. The corresponding etching rate was approximately 0.04 nm/s for the SiO2 films. Spectral data were obtained at etching times of 0 s, 50 s, 250 s, intervals of 10 s from 450 to 750 s (near the Si-SiO2 interface), and 800 s. TOF-SIMS analyses were performed in positive polarity using a pulsed Au3+ primary ion beam operated at 22 kV and 2.0 nA. The beam was rastered on a 100 × 100 µm2 surface area. Secondary ions in the mass range from 0 to 1860 atomic mass units (amu) were collected for 300 s. No charge compensation using an electron flood gun was required. The working pressure was in the 10-7 Pa range. Identification of the peaks of the secondary ion species was carried out using the mass of the peaks to an accuracy within 0.01 amu. Intensities of ion species were normalized by total ion counts. MM2 molecular mechanics calculations of the Si-O-N structures were carried out using the program package from Molecular Design Limited. RESULTS AND DISCUSSION Depth profiles of silicon cluster ions were investigated to obtain the depth resolution function of the system, that is, the response (38) Landheer, D.; Xu, D.-X.; Tao, Y.; Sproule, G. I. J. Appl. Phys. 1995, 77, 1600–1606. (39) Lucovsky, G.; Niimi, H.; Koh, K. Mater. Res. Soc. Symp. Proc. 1996, 429, 233–238. (40) Sanchez, O.; Aguilar, M. A.; Falcony, C.; Martinez-Duart, J. M.; Albella, J. M. J. Vac. Sci. Technol., A 1996, 14, 2088–2093. (41) Weidner, G.; Kru ¨ ger, D.; Weidner, M.; Grabolla, T.; Sorge, R. Microelectron. Eng. 1995, 28, 133–136. (42) Parks, C. C.; Robinson, B.; Leary, H. J., Jr.; Childs, K. D.; Coyle, G., Jr. J. Mater. Res. Soc. Symp. Proc. 1988, 105, 133–138. (43) Tobin, P. J.; Okada, Y.; Ajuria, S. A.; Lakhotia, V.; Feil, W. A.; Hedge, R. I. J. Appl. Phys. 1994, 75, 1811–1817. (44) Lee, D. R.; Parker, C. G.; Hauser, J. R.; Lucovsky, G. Mater. Res. Soc. Symp. Proc. 1995, 386, 243–248. (45) Chiba, K.; Tsuji, Y. Appl. Surf. Sci. 2008, 254, 5727–5731.


6287

Figure 1. TOF-SIMS depth profiles of (Si)n+ (n ) 1-5) silicon cluster ions of a 25-nm-thick PECVD grown SiO2 film on Si. The sputter etching rate is approximately 0.04 nm/s for the SiO2 films. The abrupt increase in intensities of the Si3+ ion species is clearly observed at the interface region between silicon and the SiO2 film.

function in the TOF-SIMS experiment. Figure 1 shows TOF-SIMS depth profiles of (Si)n+ (n ) 1-5) silicon cluster ions, that is, Si+, Si2+, Si3+, Si4+, and Si5+ ion species. The intensities of the Si+ ions emitted from the SiO2 layer show a decrease at the interface between silicon and the SiO2 film, which approximately corresponds to the sputter time of 620-650 s. Intensities of the Si+ ion appear to be almost constant in the SiO2 film, and about half-in silicon compared to those in the SiO2 film; this is mainly attributed to matrix effects associated with different sputtering efficiencies and ionization probabilities influenced by surrounding atoms.1,46,47 The Si2+ ion shows a different behavior in the depth profiles. The intensities of the Si2+ ion in the SiO2 film were low but increased at the interface from the SiO2 film to silicon. This indicates that recombination of silicon atoms, for instance, the reaction of Si+ and Si near the sputtered SiO2 surface, may occur to some extent; however, most of the Si2+ ions may be correlated to secondary ions emitted from the silicon surface. Further, intensities of both Si+ and Si2+ ions show a gradual change at the interface. On the contrary, it is to be noted that intensities of Si3+ cluster ions show an abrupt change at the interface. In this case, the intensities in the SiO2 film were negligibly small, but those from the silicon substrate appear discretely large and constant. The intensities of (Si)n+ cluster ions with the value of n more than 3 also appear to be negligibly small in the SiO2 film and increased in the silicon substrate. With an increase in n of more than 3, the intensities of (Si)n+, however, decreased and become weak. This tendency suggests that depth profiles of Si3+ ion species, corresponding to the amounts of silicon atoms in the silicon substrate, may clearly represent the shape of the interface between silicon and the SiO2 film in this experiment, which is presumed to be a step function in the original form. The depth profile of the Si3+ ion was utilized to obtain the depth-resolution function of the system. Figure 2 shows plots of the intensity changes for the Si3+ ion species near the interface (46) Andersen, C. A. Int. J. Mass Spectrom. Ion Phys. 1970, 3, 413–428. (47) Norskov, J. K.; Lundqvist, B. I. Phys. Rev. B 1979, 19, 5661–5665.

6288


between silicon and the SiO2 film. Curve-fitting using an error function was carried out, and this curve is also shown in Figure 2. Assuming a step function for the original profile at the Si-SiO2 interface, the depth resolution function of the Gaussian curve, that is, the response function of the system, was deduced. The position of the Si-SiO2 interface was determined to be about 646 s sputter time, and the standard deviation of the resolution curve was found to be 17 s with respect to the sputter time, which corresponds to 0.7 nm thickness. The full width at half-maximum (fwhm) of the depth resolution function was determined to be approximately 40 s sputter time, corresponding to 1.6 nm in distance. The intensity profile of the depth resolution function, that is, the response function of the system, is shown in Figure 2; the intensity of the curve is normalized for the integral over sputter time to unity. Deconvolution of depth profiles of nitrogen-incorporated secondary ion species at the Si-SiO2 interface was examined. Figure 3 shows depth profiles of Si3O4+ and Si3O2N+ ion species for PECVD-prepared SiO2 films on silicon. Large amounts of Si3O4+ ions were detected from the SiO2 film, and these intensities decreased at the Si-SiO2 interface. It has been reported that Si3O4+ ion species strongly correlate to the SiO2 bulk film. Some amounts of Si3O2N+ ions appeared from the SiO2 bulk film, and a significant increase of the Si3O2N+ ion species was clearly observed at the Si-SiO2 interface. It has been reported that ion species of Si3On-1N+ (n ) 2, 3, 4) observed from the SiO2 bulk film and the Si-SiO2 interface may most likely be different in the structure, whereas mixed structures may also possibly exist; the Si-O-Si(N)-O-Si+ structures are assigned to the Si3O2N+ ion species from the SiO2 bulk films, and structures with nitrogen bonded to partially oxidized silicon, Si2(O)-Si(N)-O+, are from the Si-SiO2 interface.45 The intensity ratio of Si3O2N+ to Si3O4+ ions is approximately 0.2-0.3 with an average value of 0.24 in the SiO2 bulk films. This ratio may indicate the value of the fragment ratio of Si3O2N+ to Si3O4+ from the PECVD-prepared SiO2 bulk film in this experiment. Considering this ratio of 0.24 as the correction factor for the intensity of the Si3O2N+ ion species from the Si-SiO2 interface, the depth profile of the Si3O2N+ ion

Figure 2. TOF-SIMS depth profile of Si3+ ions of a 25-nm-thick PECVD-grown SiO2 film on Si and the calculated depth resolution function. The depth resolution function (Gaussian distribution) was obtained by deconvolution of the fitted curves of the depth profile of Si3+ ions with a presumed original step function. The depth resolution function, that is the response function of a system, is normalized for the integral over sputter time to unity.

Figure 3. TOF-SIMS depth profiles of Si3O2N+ and Si3O4+ ions of a 25-nm-thick PECVD grown SiO2 film on Si. The depth profile of Si3O2N+ ions corrected after subtraction of the influence of the SiO2 bulk film is also shown. The increase in intensities of the Si3O2N+ ion species is clearly observed at the interface region between silicon and the SiO2 film.

was corrected by subtracting the correlated factor of the SiO2 bulk film; in other words, these are the values of the intensity of Si3O4+ multiplied by 0.24. The corrected depth profile of the Si3O2N+ ion species is also shown in Figure 3. Sharply increased peaks of Si3O2N+ ions, which show slow leading and sharp trailing edges, are clearly observed at the Si-SiO2 interface. The fwhm of the profile is found approximately at 60 s, which corresponds to a thickness of 2.4 nm. The asymmetrical depth profile of Si3O2N+ ions with slow leading and sharp trailing edges was examined to deconvolute it close to the original profile. The Poisson distribution of nitrogen-

incorporated species was postulated. The Poisson distribution f(x) is given as f(x) ) e-µ

µx x!

µ > 0; x ) 0, 1, 2, 3, ...

(2)

where x and µ are the discrete variable and distribution constant, respectively.48 This distribution was applied to nitrogen-incorporated structures at the Si-SiO2 interface of PECVD-prepared (48) Wadsworth, G. P.; Bryan, J. G. Applications of Probability and Random Variables, 2nd ed.; McGraw-Hill Inc.: New York, 1974.


6289

Figure 4. Depth profiles of Poisson distribution signals having various δ-time spacing and convoluted curves with the depth resolution function: (a) δ-time spacing of Poisson distribution signals of 10 s; (b) δ-time spacing of 20 s; and (c) δ-time spacing of 30 s. Different profiles of convoluted curves with respective leading and trailing slopes appear in the figure. These curves suggest that asymmetrical shapes of profiles in leading and trailing edges correlate to the δ-time spacing of original signals, which corresponds to the thickness of the δ-layer. 6290


Figure 5. Depth profiles of Si3O2N+ ions (corrected) of a 25-nmthick PECVD grown SiO2 film on Si at the Si-SiO2 interface, and the fitted curves by convoluted curves of a Poisson distribution signal having a δ-time spacing of 20 s, that is 0.8 nm thickness in terms of distance, with the depth resolution function. Close fitting is observed.

Figure 6. Molecular modeling of a part of the structure of silicon oxynitrides at the Si-SiO2 interface. Nitrogen atoms are linked to partially oxidized silicon atoms to form silicon oxynitride structures. This structure was calculated and energetically minimized using the MM2 package. It is noteworthy that the approximate size of these structures is between 0.6 and 0.8 nm.

silicon oxide films on silicon using the following assumptions: (1) The probability of existence of Si3O2N structural species in the SiO2 matrix follows a Poisson distribution function as a series of δ layers, (2) the distribution constant µ is 1, that is, the most probable layer in existence of the species is the first layer from the interface, and (3) the probability decreases from the interface to the SiO2 bulk film. Signals of the Poisson-distribution function having various δ-time spacing were postulated and convoluted with the depth-resolution function derived from depth profiles of the Si3+ ion. Figure 4 shows the original Poisson distribution signals and convoluted depth profiles of the original signals with the response function. It is to be noted that the sputter time is presented with respect to the elapsed time from the Si-SiO2 interface in Figure 4. Figure 4a shows the convoluted depth profile of the signal associated with a δ-time spacing of 10 s, which corresponds to 0.4 nm thickness in terms of distance. It is shown

that the peak maximum is shifted before the Si-SiO2 interface, and a slight asymmetrical profile with slow leading and fast trailing edges is observed. Figure 4b shows the profile of the convoluted signals associated with a δ-time spacing of 20 s (0.8 nm thick). It is clearly shown that the peak position is shifted before the interface by approximately 20 s, and notable asymmetry with slow leading and sharp trailing edges appears in the profile. Figure 4c shows the profile of the convoluted signals associated with a δ-time spacing of 30 s (1.2 nm thick). With an increase in the δ-time spacing, the peak shift and asymmetrical broadening of depth profiles increase. The experimental plots of the intensities of the Si3O2N+ ion species at the Si-SiO2 interface, corrected by subtraction of the influence of Si3O2N+ ions in the SiO2 bulk film, were fitted by curves of the postulated Poisson distribution profiles convoluted with the depth resolution function. Figure 5 shows the results of the fitting. The signal of the Poisson distribution associated with δ-time spacing of 20 s, which corresponds to 0.8 nm thickness in terms of distance, is closely fitted to both the shift of the peak position and asymmetrical depth profiles with slow leading and sharp trailing edges. This result indicates that the depth profile of the Si3O2N+ ions at the Si-SiO2 interface can be deconvoluted with the depth resolution function to the original depth profile using the Poisson distribution consisting of a series of δ-layers of 0.8 nm thickness with the distribution constant of 1. The most probable layer in existence of the nitrogen compounds is located at the first layer from the Si-SiO2 interface, and the probability of existence statistically decreases to the SiO2 bulk film. This means that nitrogen-incorporated structures may be confined within a δ-layer approximately 0.8 nm thick, and they decrease from the interface to the SiO2 bulk film followed by Poisson distribution in the original depth profile. The fwhm of the original depth profile seems to be approximately 1.6 nm thick. Figure 6 shows the molecular model of the structures of Si3O2N with the nearest neighbors at the Si-SiO2 interface. This is calculated and energetically minimized using the MM2 force field. The atomic scale of the structure is also shown in Figure 6. It is noteworthy that the size of the structure of Si3O2N with nitrogen bonded to partially oxidized silicon likely lies between 0.6 and 0.8 nm. These unit structures containing nitrogen atoms may behave as a δ-layer identified by discrete variables in the distribution. This result also indicates that nitrogen may not behave as an accumulation of atoms but rather exists as nitrogen atom-incorporated silicon structures at the Si-SiO2 interface and decreases from the interface. The approximated Poisson distribution may arise from the reaction mechanisms of SiH4 with N2O near the silicon surface due to small probability and randomness of production of the species at the interface. This structure model deconvoluted from the TOF-SIMS depth profile closely coincides with those reported by a number of researchers indicating the existence of two or three monolayers consisting of [Si-Si3-xNx]3N and [Si-Si3-xOx]3N structures at the Si-SiO2 interface.49,50 (49) Cerofolini, G. F.; Caricato, A. P.; Meda, L.; Re, N.; Sgamellotti, A. Phys. Rev. B 2000, 61, 14154–14166. (50) Oshima, M.; Kimura, K.; Ono, K.; Horiba, K.; Nakamura, K.; Kumigashira, H.; Oh, J.-H.; Niwa, M.; Usuda, K.; Hirashita, N. Appl. Surf. Sci. 2003, 216, 291–295.


6291

CONCLUSIONS Structural distributions of nitrogen-incorporated species at the interface between silicon and the SiO2 films for PECVD prepared SiO2 films have been analyzed using TOF-SIMS and the Poisson distribution model. Depth profiles of Si3+ cluster ions at the interface were used to identify the position of the interface, which is presumed originally to show an abrupt profile (a step function) of the silicon substrate, and to calculate the depth resolution function which causes beam-induced broadening effects in the original depth profiles. The intensity profile of Si3O2N+ secondary ion species at the interface was analyzed using Poisson distribution signals with a series of δ-layers. Convolution of the postulated distributions with the response function was carried out to simulate experimentally obtained depth profiles of Si3O2N+. These analyses clearly explain the shift of the position of peak maxima and asymmetrical depth profiles showing slow leading and sharp trailing edges. The calculated thickness of the δ layer with most probable existence of nitrogen-incorporated species was revealed to be approximately 0.8 nm in thickness. The probability of these fragmented species sharply decreased from the interface to the

6292


SiO2 film with a decay length of approximately 1.5 nm. These results show that nitrogen atoms are not physically incorporated as atoms into the films, but rather they bond to partially oxidized silicon atoms to form a series of δ-layers of oxynitrides and seem approximately to follow the Poisson distribution at the Si-SiO2 interface. These analytical procedures using fragmented secondary ion species in combination with deconvolution techniques may play an important role in understanding structural distribution of fragmented structures of materials in the subnanometer scale. ACKNOWLEDGMENT The author thanks Tohoku Semiconductor Co. for supplying silicon oxide films on silicon wafers. The author also thanks M. Kawamura of Tokushima Bunri University for his help with the TOF-SIMS measurements.

Received for review April 4, 2008. Accepted June 19, 2008. AC800675J

Analysis of Structural Distribution of Nitrogen-Incorporated Species at

Recommend Documents