- Email: [email protected]
H2 gas for native oxide removal
Journal of Crystal Growth 198/199 (1999) 1039—1044
Silicon epitaxial film growth on silicon substrate exposed to UV-excited NF /H gas for native oxide removal Sung Ku Kwon , Do Hyun Kim *, Jong Tae Baek Department of Chemical Engineering and Process Analysis Laboratory, Korea Advanced Institute of Science and Technology, Taejon 305-701, South Korea Semiconductor Division, Electronics and Telecommunications Research Institute, Yusong, P.O. Box 106, Taejon 305-600, South Korea
Abstract In situ UV-excited NF /H gas phase cleaning for native oxide removal and Si epitaxial film growth experiments were carried out in a load-locked reactor equipped with a UV lamp and PBN heater. The effect of composition of NF /H and UV exposure on the etching characteristics of native oxide and thermal oxide has been studied. Main etching species were identified as F, NF and HF. Hydrogen added to NF gas alleviates silicon surface pitting to result in the very smooth V surface. RMS surface roughness was as low as 0.5 A> at the ratio of H /NF of 3. Analysis of silicon eptaxial film showed the effectiveness of NF /H gas phase cleaning as the pretreatment process prior to silicon eptaxial film growth. 1999 Published by Elsevier Science B.V. All rights reserved. Keywords: Dry cleaning; UV exposure; Native oxide; NF ; Silicon epitaxy
1. Introduction Surface cleaning has been known to influence the structure of silicon epitaxial films and success of low-temperature growth processes. Various methods have been investigated for performing in situ cleaning prior to silicon epitaxial growth. These cleaning methods include thermally assisted cleaning [1], photo assisted cleaning [2], plasma assisted cleaning [3], and HF vapor cleaning [4]. Problem with conventional thermal process for native oxide removal is that it cannot desorb the * Corresponding author. Tel.: #82 42 869 3929; fax: #82 42 869 3910; e-mail: [email protected].
residual carbon contaminant and SiC precipitates are formed which are the origin of strain in the epitaxial film. Under the critical condition such as in the high temperature treatments, the induced strain relaxes via the formation of dislocations, which are fatal defects for devices. On the other hand, the quality of deposited epitaxial film is directly related to the roughness of underlying surface such that rough surfaces yields heavily twinned epitaxial films [5]. Therefore surface smoothness is indispensable to successful epitaxial film growth. In this work, in situ cleaning and silicon epitaxial film growth experiments in a load-locked reactor equipped with a Hg-grid UV lamp (BHK Inc.) and
0022-0248/99/$ — see front matter 1999 Published by Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 9 8 ) 0 1 0 7 7 - X
1040
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
PBN (pyrolytic boron nitride) heater were studied, which aims at smooth underlying surface and low temperature epitaxial film growth process. Transmission electron microscopy (TEM), Fourier transform infrared spectrometer (FT-IR), X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM) were used to characterize the cleaned surface for the surface roughness and chemistry and the deposited silicon epitaxial film for the crystalline quality.
2. Experiment For the sample preparation, thermal silicon oxide of 180 nm thick film was grown on a 4-in, p-type (1 0 0)-oriented wafer (R "22—38 ) cm) by wet 1 oxidation at 950°C in the furnace. Native (chemical) oxide was grown on the same kind of wafer as for thermal oxide after cleaning with an acetone— ethanol-deionized water in sonicating bath and drying with nitrogen blowing. The samples were cut into the pieces of 2;2 cm to be mounted on the sample holder. The cleaning experiments were carried out in a load-locked cleaning reactor equipped with a UV lamp in the reactor for the exposure to reaction gas and substrate. Prior to UV-excited NF /H cleaning for oxide removal, the wafer was treated in a UV-excited oxygen cleaning for 15 min to remove hydrocarbon molecules from the surface. Reactor chamber pressure, oxygen flow rate and UV lamp power were fixed at 2 Torr, 100 sccm and 4—5 mW/cm in the distance of 2.54 cm at the wavelength of 185 nm, respectively. Then, UV-excited NF /H gas phase cleaning experiments were carried out in the same chamber. Total flow rate, the distance between the wafer and UV lamp and cleaning time were fixed at 20 sccm, 3.5 cm and 30 min, respectively. During the gas cleaning, substrate was heated from room temperature to about 70°C by UV irradiation. Chamber pressure and UV lamp power conditions are the same as for UV/O cleaning. For the depostion of silicon epitaxial film, samples were heated from 20°C to 750°C in 5 min. H and SiH flow rates (50 and 0.5 sccm, respec tively) as well as the process pressure (350 mTorr) were established prior to heating.
Mass spectrometer (VG SX-200) was used to detect the gas phase species at various etching gas compositions. Probe position for the sampling was fixed at the height of 2 mm above the upper side of a sample surface for more effective sampling of by-product species. The sampling stream from the reactor chamber to mass spectrometer was not differentially pumped and 1/16 in stainless steel tube was used as a probe tip to minimize the effect of depletion of reactive species above the surface at the cost of missing by-product species with short life time. We also examined the removal characteristics of native oxide after gas cleaning by XPS (Specs, LHS10) using a bare wafer and a wafer with thermal oxide as references. Roughness of the surface was measured using AFM(PSI). TEM images and diffraction patterns were obtained to determine the quality of the interface and crystalline quality of deposited Si film.
3. Results and discussion To examine the growth of native oxide after wet cleaning, sample wafer was cleaned with acetone (3 min)-ethanol (3 min)-HF : D.I (1 : 10 v/v)(30 s)
Fig. 1. XPS spectra of O 1s of silicon surface as a function of exposure time in air after dipping in HF(10 : 1) for 30 s (exposure time in air: 10, 60, 120 min).
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
with deionized water rinse (3 min) after each step and was dried with nitrogen. The results of XPS analysis as a function of exposure time in air are shown in Fig. 1. After 10 min exposure in air, the existence of native oxide was confirmed in XPS spectra. Further growth of native oxide was fastly occured as the sample was exposed in air. This suggests that the removal of native oxide by HF wet cleaning is not complete to prevent the regrowth of native oxide. Fig. 2 shows XPS spectra of (a) Si 2p, (b) O 1s, and (c) F 1s peaks obtained from a bare silicon wafer, a processed wafer and a wafer with thermal oxide surface. XPS spectra of Si 2p and O 1s obtained from a processed sample show the effective removal of native oxide by UV-excited NF /H gas cleaning. Weak intensity of O 1s and Si 2p
1041
obtained from processed wafer is thought to be caused by air exposure during transportation to analyzer chamber. F 1s peak in Fig. 2c is actually a convolution of two peaks. One of the peaks is associated with a weakly bound state of fluorine, while the other peak is due to a strongly bound chemisorbed state of fluorine. F 1s peak at 687.6 eV of processed wafer is associated with a weakly bound state and that of thermal oxide is associated with a strongly bound state of fluorine. The intensity of F 1s peak and the degree of hydrogen termination mainly depend on process conditions, density of step edges and composition of H /NF in reacting gas. To investigate the effect of UV irradiation, we performed residual gas analysis (RGA) during the etching of silicon dioxide by sequentially turning
Fig. 2. XPS spectra of (a) Si 2p, (b) O 1s and (c) F 1s peaks on a bare wafer, a processed wafer, and a wafer with thermal oxide (pressure"2 Torr, NF /H flow rate"10/10 sccm).
1042
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
on and off the UV lamp. Fig. 3 shows the results of residual gas analysis during silicon dioxide etching. The gas species in the gas mixture were analyzed by mass spectrometer to be F, HF, NO , NF and SiF at mass numbers of 19, 20, 46, 52 and 85, respectively. These species are results of the reaction between NF and H to form NF , HF and V additional fluorine atoms. UV irradiation enhances oxide etching as can be seen during the turning-on period in Fig. 3, which shows the consumption of etching reactants and the generation of etching products. Active species for SiO etching were iden tified as F, HF and NF , while NO and SiF were V produced as by-product.
Fig. 3. Residual gas analysis during silicon dioxide etching in UV-excited NF /H gas with UV power turned on and off (pressure"2 Torr, NF /H flow rate"10/10 sccm).
Etch rate of SiO /Si was indirectly obtained by measuring the SiF partial pressure by mass spec trometer. Although the etch rates of thermal oxide and silicon decreased with the increasing ratio of H /NF flow rate, the etch selectivity of SiO /Si increased to 1.8. With native oxide we expect higher value of etch selectivity of SiO /Si since native oxide is known to be etched much easier than thermal oxide [6]. Partial pressure variation of H and HF was measured by mass spectrometer as a function of H flow rate during oxide etching. Partial pressure of F was expected to decrease with the increase of the ratio of H /NF flow rate. In Fig. 4, however, it did not decrease as much as expected. Partial pressure of HF increased with flow rate ratio and shows maximum at the flow rate ratio of about 1 (10 sccm), as expected. The reason why the partial presssure of F did not decrease as much as expected can be speculated as the cracking of HF in the mass spectrometer. RMS (root mean square) roughness of processed wafer surface was measured by AFM and shown in Fig. 5 as a function of H flow rate. The RMS roughness on silicon surface after gas phase cleaning can be divided into three regimes by H con tents in gas mixture, values of RMS roughness and surface topography. Corresponding AFM images in each regime are shown together. The etching of native oxides in UV-excited NF gas causes pitting on silicon surface through excessive etching of silicon by fluorine. Hydrogen scavenges fluorine to
Fig. 4. (a) F and (b) HF partial pressure variation measured by mass spectrometer as a function of H flow rate during oxide etching (total flow rate"20 sccm, pressure"2 Torr).
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
1043
Fig. 5. RMS roughness and AFM surface images of processed wafer surfaces as a function of H flow rate (pressure"2.0 Torr, total flow rate"20 sccm).
form HF and prevents excessive etching of silicon [6]. Silicon surface pitting was not observed when content of H in the etching gas mixture was 75% or higher, where RMS surface roughness was measured as 0.5 A> . We have taken the TEM image of silicon epitaxial film grown after NF /H gas phase cleaning. Interfacial smoothness is very good because oxide removal was carried out under conditions corresponding to regime (III) in Fig. 6. However, the oxide layer is seen to be completely removed. Diffraction pattern of epitaxial layer using SAD (selected area diffraction) technique confirmed crystallinity of deposited layer. SAD pattern of silicon epi-layer and substrate was inserted in Fig. 6.
4. Conclusions In situ UV excited NF /H gas phase cleaning for native oxide removal and Si epitaxial film growth experiments were carried out in a loadlocked reactor equipped with a UV lamp and PBN heater. The effects of composition of NF /H and UV exposure on the etching characteristics of native oxide and thermal oxide have been studied. With UV exposure and the optimal composition of NF /H , native oxide was effectively removed. Also, surface hydrogen termination was as good as wet HF cleaned silicon surface. According to mass spectrometry analysis, main etching species were F, NF and HF while by-products were NO and V
1044
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
SiF . Hydrogen added to NF gas alleviated silicon surface pitting although it decreased the etch rate of thermal oxide. RMS surface roughness of the processed wafer measured by AFM was as low as 0.5 A> when the ratio of H /NF is 3 or higher. Analysis of deposited silicon epitaxial film showed good crystalline quality of film and effectiveness of NF /H gas cleaning process.
References [1] T. Yamaraki, N. Miyata, T. Aoyama, T. Ito, J. Electrochem. Soc. 139 (1992) 1175. [2] T. Aoyama, T. Yamazaki, T. Ito, J. Electrochem. Soc. 140 (1993) 1704. [3] R.P.H. Chang, C.C.C. Chang, S. Darack, J. Vac. Sci. Technol. 20 (1982) 45. [4] M. Wong, M.M. Moslehi, D.W. Reed, J. Electromchem. Soc. 138 (1991) 1799. [5] P.P. Buaud, Y.Z. Hu, L. Spanos, E.A. Irene, K.N. Christensen, D. Venables, D.M. Maher, J. Vac. Sci. Technol. B 13 (1995) 1442. [6] K. Torek, J. Ruzyllo, Proc. 2nd Int. Symp. on Cleaning Technology in Semiconductor Device Manufacturing, 92-2, 1992, p. 80. Fig. 6. TEM image of epitaxial film (maginification"50 000, deposition condition: SiH flow rate"50 sccm, H flow rate" 0.5 sccm, pressure"200 mTorr, temperature"750°C).
Silicon epitaxial film growth on silicon substrate exposed to UV-excited NF /H gas for native oxide removal Sung Ku Kwon , Do Hyun Kim *, Jong Tae Baek Department of Chemical Engineering and Process Analysis Laboratory, Korea Advanced Institute of Science and Technology, Taejon 305-701, South Korea Semiconductor Division, Electronics and Telecommunications Research Institute, Yusong, P.O. Box 106, Taejon 305-600, South Korea
Abstract In situ UV-excited NF /H gas phase cleaning for native oxide removal and Si epitaxial film growth experiments were carried out in a load-locked reactor equipped with a UV lamp and PBN heater. The effect of composition of NF /H and UV exposure on the etching characteristics of native oxide and thermal oxide has been studied. Main etching species were identified as F, NF and HF. Hydrogen added to NF gas alleviates silicon surface pitting to result in the very smooth V surface. RMS surface roughness was as low as 0.5 A> at the ratio of H /NF of 3. Analysis of silicon eptaxial film showed the effectiveness of NF /H gas phase cleaning as the pretreatment process prior to silicon eptaxial film growth. 1999 Published by Elsevier Science B.V. All rights reserved. Keywords: Dry cleaning; UV exposure; Native oxide; NF ; Silicon epitaxy
1. Introduction Surface cleaning has been known to influence the structure of silicon epitaxial films and success of low-temperature growth processes. Various methods have been investigated for performing in situ cleaning prior to silicon epitaxial growth. These cleaning methods include thermally assisted cleaning [1], photo assisted cleaning [2], plasma assisted cleaning [3], and HF vapor cleaning [4]. Problem with conventional thermal process for native oxide removal is that it cannot desorb the * Corresponding author. Tel.: #82 42 869 3929; fax: #82 42 869 3910; e-mail: [email protected].
residual carbon contaminant and SiC precipitates are formed which are the origin of strain in the epitaxial film. Under the critical condition such as in the high temperature treatments, the induced strain relaxes via the formation of dislocations, which are fatal defects for devices. On the other hand, the quality of deposited epitaxial film is directly related to the roughness of underlying surface such that rough surfaces yields heavily twinned epitaxial films [5]. Therefore surface smoothness is indispensable to successful epitaxial film growth. In this work, in situ cleaning and silicon epitaxial film growth experiments in a load-locked reactor equipped with a Hg-grid UV lamp (BHK Inc.) and
0022-0248/99/$ — see front matter 1999 Published by Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 9 8 ) 0 1 0 7 7 - X
1040
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
PBN (pyrolytic boron nitride) heater were studied, which aims at smooth underlying surface and low temperature epitaxial film growth process. Transmission electron microscopy (TEM), Fourier transform infrared spectrometer (FT-IR), X-ray photoelectron spectroscopy (XPS) and atomic force microscopy (AFM) were used to characterize the cleaned surface for the surface roughness and chemistry and the deposited silicon epitaxial film for the crystalline quality.
2. Experiment For the sample preparation, thermal silicon oxide of 180 nm thick film was grown on a 4-in, p-type (1 0 0)-oriented wafer (R "22—38 ) cm) by wet 1 oxidation at 950°C in the furnace. Native (chemical) oxide was grown on the same kind of wafer as for thermal oxide after cleaning with an acetone— ethanol-deionized water in sonicating bath and drying with nitrogen blowing. The samples were cut into the pieces of 2;2 cm to be mounted on the sample holder. The cleaning experiments were carried out in a load-locked cleaning reactor equipped with a UV lamp in the reactor for the exposure to reaction gas and substrate. Prior to UV-excited NF /H cleaning for oxide removal, the wafer was treated in a UV-excited oxygen cleaning for 15 min to remove hydrocarbon molecules from the surface. Reactor chamber pressure, oxygen flow rate and UV lamp power were fixed at 2 Torr, 100 sccm and 4—5 mW/cm in the distance of 2.54 cm at the wavelength of 185 nm, respectively. Then, UV-excited NF /H gas phase cleaning experiments were carried out in the same chamber. Total flow rate, the distance between the wafer and UV lamp and cleaning time were fixed at 20 sccm, 3.5 cm and 30 min, respectively. During the gas cleaning, substrate was heated from room temperature to about 70°C by UV irradiation. Chamber pressure and UV lamp power conditions are the same as for UV/O cleaning. For the depostion of silicon epitaxial film, samples were heated from 20°C to 750°C in 5 min. H and SiH flow rates (50 and 0.5 sccm, respec tively) as well as the process pressure (350 mTorr) were established prior to heating.
Mass spectrometer (VG SX-200) was used to detect the gas phase species at various etching gas compositions. Probe position for the sampling was fixed at the height of 2 mm above the upper side of a sample surface for more effective sampling of by-product species. The sampling stream from the reactor chamber to mass spectrometer was not differentially pumped and 1/16 in stainless steel tube was used as a probe tip to minimize the effect of depletion of reactive species above the surface at the cost of missing by-product species with short life time. We also examined the removal characteristics of native oxide after gas cleaning by XPS (Specs, LHS10) using a bare wafer and a wafer with thermal oxide as references. Roughness of the surface was measured using AFM(PSI). TEM images and diffraction patterns were obtained to determine the quality of the interface and crystalline quality of deposited Si film.
3. Results and discussion To examine the growth of native oxide after wet cleaning, sample wafer was cleaned with acetone (3 min)-ethanol (3 min)-HF : D.I (1 : 10 v/v)(30 s)
Fig. 1. XPS spectra of O 1s of silicon surface as a function of exposure time in air after dipping in HF(10 : 1) for 30 s (exposure time in air: 10, 60, 120 min).
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
with deionized water rinse (3 min) after each step and was dried with nitrogen. The results of XPS analysis as a function of exposure time in air are shown in Fig. 1. After 10 min exposure in air, the existence of native oxide was confirmed in XPS spectra. Further growth of native oxide was fastly occured as the sample was exposed in air. This suggests that the removal of native oxide by HF wet cleaning is not complete to prevent the regrowth of native oxide. Fig. 2 shows XPS spectra of (a) Si 2p, (b) O 1s, and (c) F 1s peaks obtained from a bare silicon wafer, a processed wafer and a wafer with thermal oxide surface. XPS spectra of Si 2p and O 1s obtained from a processed sample show the effective removal of native oxide by UV-excited NF /H gas cleaning. Weak intensity of O 1s and Si 2p
1041
obtained from processed wafer is thought to be caused by air exposure during transportation to analyzer chamber. F 1s peak in Fig. 2c is actually a convolution of two peaks. One of the peaks is associated with a weakly bound state of fluorine, while the other peak is due to a strongly bound chemisorbed state of fluorine. F 1s peak at 687.6 eV of processed wafer is associated with a weakly bound state and that of thermal oxide is associated with a strongly bound state of fluorine. The intensity of F 1s peak and the degree of hydrogen termination mainly depend on process conditions, density of step edges and composition of H /NF in reacting gas. To investigate the effect of UV irradiation, we performed residual gas analysis (RGA) during the etching of silicon dioxide by sequentially turning
Fig. 2. XPS spectra of (a) Si 2p, (b) O 1s and (c) F 1s peaks on a bare wafer, a processed wafer, and a wafer with thermal oxide (pressure"2 Torr, NF /H flow rate"10/10 sccm).
1042
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
on and off the UV lamp. Fig. 3 shows the results of residual gas analysis during silicon dioxide etching. The gas species in the gas mixture were analyzed by mass spectrometer to be F, HF, NO , NF and SiF at mass numbers of 19, 20, 46, 52 and 85, respectively. These species are results of the reaction between NF and H to form NF , HF and V additional fluorine atoms. UV irradiation enhances oxide etching as can be seen during the turning-on period in Fig. 3, which shows the consumption of etching reactants and the generation of etching products. Active species for SiO etching were iden tified as F, HF and NF , while NO and SiF were V produced as by-product.
Fig. 3. Residual gas analysis during silicon dioxide etching in UV-excited NF /H gas with UV power turned on and off (pressure"2 Torr, NF /H flow rate"10/10 sccm).
Etch rate of SiO /Si was indirectly obtained by measuring the SiF partial pressure by mass spec trometer. Although the etch rates of thermal oxide and silicon decreased with the increasing ratio of H /NF flow rate, the etch selectivity of SiO /Si increased to 1.8. With native oxide we expect higher value of etch selectivity of SiO /Si since native oxide is known to be etched much easier than thermal oxide [6]. Partial pressure variation of H and HF was measured by mass spectrometer as a function of H flow rate during oxide etching. Partial pressure of F was expected to decrease with the increase of the ratio of H /NF flow rate. In Fig. 4, however, it did not decrease as much as expected. Partial pressure of HF increased with flow rate ratio and shows maximum at the flow rate ratio of about 1 (10 sccm), as expected. The reason why the partial presssure of F did not decrease as much as expected can be speculated as the cracking of HF in the mass spectrometer. RMS (root mean square) roughness of processed wafer surface was measured by AFM and shown in Fig. 5 as a function of H flow rate. The RMS roughness on silicon surface after gas phase cleaning can be divided into three regimes by H con tents in gas mixture, values of RMS roughness and surface topography. Corresponding AFM images in each regime are shown together. The etching of native oxides in UV-excited NF gas causes pitting on silicon surface through excessive etching of silicon by fluorine. Hydrogen scavenges fluorine to
Fig. 4. (a) F and (b) HF partial pressure variation measured by mass spectrometer as a function of H flow rate during oxide etching (total flow rate"20 sccm, pressure"2 Torr).
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
1043
Fig. 5. RMS roughness and AFM surface images of processed wafer surfaces as a function of H flow rate (pressure"2.0 Torr, total flow rate"20 sccm).
form HF and prevents excessive etching of silicon [6]. Silicon surface pitting was not observed when content of H in the etching gas mixture was 75% or higher, where RMS surface roughness was measured as 0.5 A> . We have taken the TEM image of silicon epitaxial film grown after NF /H gas phase cleaning. Interfacial smoothness is very good because oxide removal was carried out under conditions corresponding to regime (III) in Fig. 6. However, the oxide layer is seen to be completely removed. Diffraction pattern of epitaxial layer using SAD (selected area diffraction) technique confirmed crystallinity of deposited layer. SAD pattern of silicon epi-layer and substrate was inserted in Fig. 6.
4. Conclusions In situ UV excited NF /H gas phase cleaning for native oxide removal and Si epitaxial film growth experiments were carried out in a loadlocked reactor equipped with a UV lamp and PBN heater. The effects of composition of NF /H and UV exposure on the etching characteristics of native oxide and thermal oxide have been studied. With UV exposure and the optimal composition of NF /H , native oxide was effectively removed. Also, surface hydrogen termination was as good as wet HF cleaned silicon surface. According to mass spectrometry analysis, main etching species were F, NF and HF while by-products were NO and V
1044
S.K. Kwon et al. / Journal of Crystal Growth 198/199 (1999) 1039–1044
SiF . Hydrogen added to NF gas alleviated silicon surface pitting although it decreased the etch rate of thermal oxide. RMS surface roughness of the processed wafer measured by AFM was as low as 0.5 A> when the ratio of H /NF is 3 or higher. Analysis of deposited silicon epitaxial film showed good crystalline quality of film and effectiveness of NF /H gas cleaning process.
References [1] T. Yamaraki, N. Miyata, T. Aoyama, T. Ito, J. Electrochem. Soc. 139 (1992) 1175. [2] T. Aoyama, T. Yamazaki, T. Ito, J. Electrochem. Soc. 140 (1993) 1704. [3] R.P.H. Chang, C.C.C. Chang, S. Darack, J. Vac. Sci. Technol. 20 (1982) 45. [4] M. Wong, M.M. Moslehi, D.W. Reed, J. Electromchem. Soc. 138 (1991) 1799. [5] P.P. Buaud, Y.Z. Hu, L. Spanos, E.A. Irene, K.N. Christensen, D. Venables, D.M. Maher, J. Vac. Sci. Technol. B 13 (1995) 1442. [6] K. Torek, J. Ruzyllo, Proc. 2nd Int. Symp. on Cleaning Technology in Semiconductor Device Manufacturing, 92-2, 1992, p. 80. Fig. 6. TEM image of epitaxial film (maginification"50 000, deposition condition: SiH flow rate"50 sccm, H flow rate" 0.5 sccm, pressure"200 mTorr, temperature"750°C).