- Email: [email protected]
LEED analysis of solid phase epitaxy of Si
Surface Science 130 (1983) L325-L328 North-Holland Publishing Company
L325
SURFACE SCIENCE LETTERS LEED ANALYSIS OF SOLID P H A S E EPITAXY O F Si
V.V. KOROBTSOV, V.G. ZAVODINSKII and A.V. ZOTOV Institute of Automation and Control Processes, Academy of Sciences of the USSR, Vladivostok 690032, USSR Received 13 April 1983
The LEED technique has been used to study the solid phase epitaxial growth of silicon. The crystalline quality of the film surface, the growth rate and the activation energy have been determined. A comparison of these results with those obtained by other authors has been made.
The interest to the process of solid phase epitaxy (SPE) arises both from the scientific viewpoint to study the fundamental physical mechanisms involved and from its potential as a low-temperature film growth method in device fabrication. The main experimental techniques nowadays applied to the study of SPE are Rutherford backscattering (RBS) and transmission electron microscopy (TEM). The RBS technique gives reasonably good quantitative information on the motion of amorphous-to-crystalline interfaces and, thus, is suited for studying crystallization kinetics [1-7]. TEM analysis is usually used for studying crystal defects in annealed layers [8]. However, Drosd and Washburn [9] using the original technique of observation spread TEM to the study of SPE kinetics. Unfortunately, both RBS and TEM can hardly be applied for in situ investigations in ultra-high vacuum. Vacuum conditions are sufficiently important in the study of SPE of deposited amorphous films since they can absorb significant amounts of ambient gases through the interconnected network of voids [6,7,10]. In this paper we report on the study of SPE by low energy electron diffraction (LEED) which provides the continuous control of the sample surface structure during annealing in ultra-high vacuum. The experimental procedure was as follows: Amorphous silicon films were deposited in ultra-high vacuum (p ~< 6 x 10 -1° Torr) onto atomically-clean crystalline silicon substrates. The amorphous state of the sample surface was indicated by a high background level and complete absence of all diffraction spots on the LEED pattern. After deposition the films were annealed by passing current through the sample. The epitaxial crystallization of amorphous silicon films on monocrystalline silicon substrates is known to proceed by the movement of the 0039-6028/83/0000-0000/$03.00 © 1983 North-Holland
L326
P: b< Korobtsoe et M. / LEED analysis of SPE of Si
amorphous-to-crystalline interface from tile substrate towards the surface [1,2,4]. Since LEED provides information concerning the structure of tile surface, the appearance of diffraction spots on the LEED pattern corresponds to the moment when the crystallization front reaches the surface. Therefore, the time interval t 0 from the beginning of the annealing up to the appearance of the spotty LEED pattern characterizes the rate of crystallization. The kinetics of the crystallization front motion can be studied by measuring t~ for films of different thicknesses d. As an example, fig. 1 represents the thickness dependence of t o for the films deposited onto Si(100) at 100°C and annealed at 540°C. Linear dependence corresponds to the movement of an amorphous-tocrystalline interface with constant rate. The growth rate derived from the plot slope is 120 A/rain. The temperature dependence of the growth rate studied at temperatures ranging from 500 to 560°C is represented by an Arrhenius-type expression with activation energy of 3.1 _+ 0.3 eV. To compare these data with other data, activation energies and regrowth rates obtained by different authors for Si(I00) samples are listed in table 1. One can see that there is a certain discrepancy both in growth rates and in activation energies. However. we speculate that it is connected rather with different conditions of amorphous film preparation and annealing than with peculiarities of research techniques. It is seen in fig. 1 that the plot of to(d) does not pass zero of the coordinate system but crosses the thickness axis at d = 100 ,~.. The LEED observation of films thinner than 100 A reveals that the diffraction spots fade with increasing thickness of the deposited film and disappear completely at d = 100 A. This is explained by the partial ordering of the deposited atoms on the initial stages of deposition at the substrate temperature (100°C) and deposition rate (100
10
£i(100)o
10
/
'--" 08
'2
0
Z
C~
o
>- 0.4 ~
500 FILM
1000 THICKNESS
1500
B
0.2 0 0 °--~
__._e_..~.lO. . . . . 2~0 {~-----'~-~-80
(.'~)
Fig. I. Fihn thickness dependence of t, for Si(100) sample annealed at 5400( ". Fig. 2. Intensity ratio 1 / 1 o versus annealing time for Si(100) sample annealed at 540°( ' (curve A) and Si(111) sample annealed at 600°C (curve B). The thickness of the deposited film is 750 /i, for Si(100) and 800 A for Si(111),
L327
V. IL Korobtsov et a L / LEED analysis of SPE of Si
Table 1 Growth rates and activation energies for Si(100) obtained by different authors Experimental technique
Amorphous film preparation
Growth rate (,~/min)
Activation energy (eV)
Ref.
RBS
Implantation of 2SSi ions
90 at 550°C
2.4
[1,2]
80
2.85
[51
Deposition
at 550°C 80 at 625°C
1.2
[7]
70 TEM LEED
Implantation of 3tp ions Deposition
at 550°C 230 at 550°C 120 at 5400C
[41 2.9
[91
3.1
This work
,~/min) used in the experiment. That is, the initial amorphous-to-crystalline interface is not abrupt and there is a transition layer of mixed structure between crystalline substrate and "purely" amorphous film. The crystallization originated on the upper boundary of the transition layer determines the shape of the observed thickness dependence of t o. The perfect monocrystalline regions of the surface contribute to the intensities of the LEED spots, while the disordered regions raise the background level. Therefore the brightness of the LEED pattern indicates the crystalline quality of the surface. Unfortunately the quantitative relationship between the spot intensity and the type and density of the surface defects is not yet available [11]. However, the surface perfection can be qualitatively represented by the intensity ratio 1/lo, where I is the spot intensity of the LEED pattern from the film and I 0 the same spot intensity from the perfect substrate. A perfect crystalline surface corresponds to 1/10 = 1, and a completely disordered (amorphous or finely polycrystalline) surface to I / I o = 0. The intensities are measured above the background level. Fig. 2 shows the intensity ratio I / I o versus annealing time for films deposited onto Si(100) (curve A) and Si(111) (curve B) and annealed at 540 and 600°C, respectively. One can see in fig. 2 that after the appearance of the spotty LEED pattern the spot intensity grows and after a certain time it achieves the level of I / I o = 1 in the case of a Si(100) sample, while for Si(lll) the achieved level is considerably lower ( I / I o-~ 0.2). This means that the structural quality of the annealed film on Si(100) is comparable with that of the substrate, while the (111) sample shows high residual disorder. The latter is confirmed by TEM analysis [12]. It should be pointed o u t that the observed drastic difference in the perfection of the
1.32~
V.V. Korobtsov et al. / LEED analysis of SPE of Si
(111) and (l()0) samples after heat treatment agrees with the data obtained by other techniques [2]. Thus, it has been shown that LEED analysis can be successfully applied to the quantitative study of the amorphous-to-crystalline interface motion and to a qualitative estimate of the annealed layer perfection.
References [1] [2] [3] [4] [5] [6] [7] [8] [91 [10l [11] [12]
L. Csepregi, J.W. Mayer and T.W. Sigmon, Phys. Letters 54 (1975) 157. S.S. Lau, J. Vacuum Sci. Technol. 15 (1978) 165,5. M. yon Alhnen, S.S. Lau, W.F. Tseng and J.W. Mayer, Appl. Phys. Letters 35 (1979) 280. L.S. Hung, S.S, Lau, M. yon Allmen, J.W. Mayer, B.M. Ullrich, J.E. Baker, P. Williams alad W.F. Tseng, Appl. Phys. Letters 37 (1980) 909. A. Lietoila, A. Wakita, T.W. Sigmon and J.F. Gibbons, J. Appl. Phys. 53 (1982) 4399. J.C. Bean and J.M. Poate, Appl. Phys. Letters 36 (1980)59. G. Foti, J.C. Bean, J.M. Poate and C.W. Magee, Appl. Phys. Letters 36 (1980) 840. L.D. Glowinski, K.N. Tu and P.S. Ho, Appl. Phys. Letters 28 (1976) 312. D. Drosd and J. Washburn, J. Appl. Phys. 51 (1980) 4106. ('.W. Magee, J.C. Bean, G. Foti and J.M. Poate, Thin Solid Films 8I (1981) 1. M. Henzler, in: FestkOrperprobleine (Advances in Solid State Physics), Vol. 19, Ed. J. Treusch (Vieweg, Braunschweig, 1979) p. 193. V.V. Korobtsov, V.G. Zavodinskii and A.V. Zotov, Phys. Status Solidi (a~ 72 (1982) 391.
L325
SURFACE SCIENCE LETTERS LEED ANALYSIS OF SOLID P H A S E EPITAXY O F Si
V.V. KOROBTSOV, V.G. ZAVODINSKII and A.V. ZOTOV Institute of Automation and Control Processes, Academy of Sciences of the USSR, Vladivostok 690032, USSR Received 13 April 1983
The LEED technique has been used to study the solid phase epitaxial growth of silicon. The crystalline quality of the film surface, the growth rate and the activation energy have been determined. A comparison of these results with those obtained by other authors has been made.
The interest to the process of solid phase epitaxy (SPE) arises both from the scientific viewpoint to study the fundamental physical mechanisms involved and from its potential as a low-temperature film growth method in device fabrication. The main experimental techniques nowadays applied to the study of SPE are Rutherford backscattering (RBS) and transmission electron microscopy (TEM). The RBS technique gives reasonably good quantitative information on the motion of amorphous-to-crystalline interfaces and, thus, is suited for studying crystallization kinetics [1-7]. TEM analysis is usually used for studying crystal defects in annealed layers [8]. However, Drosd and Washburn [9] using the original technique of observation spread TEM to the study of SPE kinetics. Unfortunately, both RBS and TEM can hardly be applied for in situ investigations in ultra-high vacuum. Vacuum conditions are sufficiently important in the study of SPE of deposited amorphous films since they can absorb significant amounts of ambient gases through the interconnected network of voids [6,7,10]. In this paper we report on the study of SPE by low energy electron diffraction (LEED) which provides the continuous control of the sample surface structure during annealing in ultra-high vacuum. The experimental procedure was as follows: Amorphous silicon films were deposited in ultra-high vacuum (p ~< 6 x 10 -1° Torr) onto atomically-clean crystalline silicon substrates. The amorphous state of the sample surface was indicated by a high background level and complete absence of all diffraction spots on the LEED pattern. After deposition the films were annealed by passing current through the sample. The epitaxial crystallization of amorphous silicon films on monocrystalline silicon substrates is known to proceed by the movement of the 0039-6028/83/0000-0000/$03.00 © 1983 North-Holland
L326
P: b< Korobtsoe et M. / LEED analysis of SPE of Si
amorphous-to-crystalline interface from tile substrate towards the surface [1,2,4]. Since LEED provides information concerning the structure of tile surface, the appearance of diffraction spots on the LEED pattern corresponds to the moment when the crystallization front reaches the surface. Therefore, the time interval t 0 from the beginning of the annealing up to the appearance of the spotty LEED pattern characterizes the rate of crystallization. The kinetics of the crystallization front motion can be studied by measuring t~ for films of different thicknesses d. As an example, fig. 1 represents the thickness dependence of t o for the films deposited onto Si(100) at 100°C and annealed at 540°C. Linear dependence corresponds to the movement of an amorphous-tocrystalline interface with constant rate. The growth rate derived from the plot slope is 120 A/rain. The temperature dependence of the growth rate studied at temperatures ranging from 500 to 560°C is represented by an Arrhenius-type expression with activation energy of 3.1 _+ 0.3 eV. To compare these data with other data, activation energies and regrowth rates obtained by different authors for Si(I00) samples are listed in table 1. One can see that there is a certain discrepancy both in growth rates and in activation energies. However. we speculate that it is connected rather with different conditions of amorphous film preparation and annealing than with peculiarities of research techniques. It is seen in fig. 1 that the plot of to(d) does not pass zero of the coordinate system but crosses the thickness axis at d = 100 ,~.. The LEED observation of films thinner than 100 A reveals that the diffraction spots fade with increasing thickness of the deposited film and disappear completely at d = 100 A. This is explained by the partial ordering of the deposited atoms on the initial stages of deposition at the substrate temperature (100°C) and deposition rate (100
10
£i(100)o
10
/
'--" 08
'2
0
Z
C~
o
>- 0.4 ~
500 FILM
1000 THICKNESS
1500
B
0.2 0 0 °--~
__._e_..~.lO. . . . . 2~0 {~-----'~-~-80
(.'~)
Fig. I. Fihn thickness dependence of t, for Si(100) sample annealed at 5400( ". Fig. 2. Intensity ratio 1 / 1 o versus annealing time for Si(100) sample annealed at 540°( ' (curve A) and Si(111) sample annealed at 600°C (curve B). The thickness of the deposited film is 750 /i, for Si(100) and 800 A for Si(111),
L327
V. IL Korobtsov et a L / LEED analysis of SPE of Si
Table 1 Growth rates and activation energies for Si(100) obtained by different authors Experimental technique
Amorphous film preparation
Growth rate (,~/min)
Activation energy (eV)
Ref.
RBS
Implantation of 2SSi ions
90 at 550°C
2.4
[1,2]
80
2.85
[51
Deposition
at 550°C 80 at 625°C
1.2
[7]
70 TEM LEED
Implantation of 3tp ions Deposition
at 550°C 230 at 550°C 120 at 5400C
[41 2.9
[91
3.1
This work
,~/min) used in the experiment. That is, the initial amorphous-to-crystalline interface is not abrupt and there is a transition layer of mixed structure between crystalline substrate and "purely" amorphous film. The crystallization originated on the upper boundary of the transition layer determines the shape of the observed thickness dependence of t o. The perfect monocrystalline regions of the surface contribute to the intensities of the LEED spots, while the disordered regions raise the background level. Therefore the brightness of the LEED pattern indicates the crystalline quality of the surface. Unfortunately the quantitative relationship between the spot intensity and the type and density of the surface defects is not yet available [11]. However, the surface perfection can be qualitatively represented by the intensity ratio 1/lo, where I is the spot intensity of the LEED pattern from the film and I 0 the same spot intensity from the perfect substrate. A perfect crystalline surface corresponds to 1/10 = 1, and a completely disordered (amorphous or finely polycrystalline) surface to I / I o = 0. The intensities are measured above the background level. Fig. 2 shows the intensity ratio I / I o versus annealing time for films deposited onto Si(100) (curve A) and Si(111) (curve B) and annealed at 540 and 600°C, respectively. One can see in fig. 2 that after the appearance of the spotty LEED pattern the spot intensity grows and after a certain time it achieves the level of I / I o = 1 in the case of a Si(100) sample, while for Si(lll) the achieved level is considerably lower ( I / I o-~ 0.2). This means that the structural quality of the annealed film on Si(100) is comparable with that of the substrate, while the (111) sample shows high residual disorder. The latter is confirmed by TEM analysis [12]. It should be pointed o u t that the observed drastic difference in the perfection of the
1.32~
V.V. Korobtsov et al. / LEED analysis of SPE of Si
(111) and (l()0) samples after heat treatment agrees with the data obtained by other techniques [2]. Thus, it has been shown that LEED analysis can be successfully applied to the quantitative study of the amorphous-to-crystalline interface motion and to a qualitative estimate of the annealed layer perfection.
References [1] [2] [3] [4] [5] [6] [7] [8] [91 [10l [11] [12]
L. Csepregi, J.W. Mayer and T.W. Sigmon, Phys. Letters 54 (1975) 157. S.S. Lau, J. Vacuum Sci. Technol. 15 (1978) 165,5. M. yon Alhnen, S.S. Lau, W.F. Tseng and J.W. Mayer, Appl. Phys. Letters 35 (1979) 280. L.S. Hung, S.S, Lau, M. yon Allmen, J.W. Mayer, B.M. Ullrich, J.E. Baker, P. Williams alad W.F. Tseng, Appl. Phys. Letters 37 (1980) 909. A. Lietoila, A. Wakita, T.W. Sigmon and J.F. Gibbons, J. Appl. Phys. 53 (1982) 4399. J.C. Bean and J.M. Poate, Appl. Phys. Letters 36 (1980)59. G. Foti, J.C. Bean, J.M. Poate and C.W. Magee, Appl. Phys. Letters 36 (1980) 840. L.D. Glowinski, K.N. Tu and P.S. Ho, Appl. Phys. Letters 28 (1976) 312. D. Drosd and J. Washburn, J. Appl. Phys. 51 (1980) 4106. ('.W. Magee, J.C. Bean, G. Foti and J.M. Poate, Thin Solid Films 8I (1981) 1. M. Henzler, in: FestkOrperprobleine (Advances in Solid State Physics), Vol. 19, Ed. J. Treusch (Vieweg, Braunschweig, 1979) p. 193. V.V. Korobtsov, V.G. Zavodinskii and A.V. Zotov, Phys. Status Solidi (a~ 72 (1982) 391.