- Email: [email protected]
Behavior of monolayer holes on MBE grown GaAs surfaces during annealing revealed by in situ scanning electron microscopy
,. . . . . . . . C R Y S T A L GROWTH
ELSEVIER
Journal of Crystal Growth 164 (1996) 88-91
Behavior of monolayer holes on MBE grown GaAs surfaces during annealing revealed by in situ scanning electron microscopy N. Inoue
a, *
y . H o m m a b j. O s a k a c T. A r a k i a
a Research Institute for Advanced Science and Technology, Osaka Prefecture University, 1-2, Gakuencho, Sakai, Osaka 593, Japan b NTT Interdisciplinary Research Laboratories, 3-9-11, Musashino, Tokyo 180, Japan c NTT LSI Laboratories, 3-1, Morinosato-Wakamiya, Atsugi, Kanagawa 243-01, Japan
Abstract
The behavior of monolayer-deep holes during annealing after molecular beam epitaxy growth was observed in situ by scanning electron microscopy. Most of the small holes formed by coalescence of islands during growth shrink and disappear as soon as the islands disappear. However some holes are left unremoved and grow and coalesce with each other to form large holes. They grow in the [110] direction but shrink in the [170] direction. They finally shrink and disappear at about 10 min. It is shown that annihilation of holes plays an important role in surface smoothing during annealing after growth. The anisotropic behavior is attributed to the preference of detachment of Ga atoms from Ga-terminated A-steps and attachment to As-terminated B-steps.
1. I n t r o d u c t i o n
It is well known that annealing after MBE or CBE growth smoothes the surface [1]. This behavior was successfully utilized to obtain smooth interfaces in quantum wells [2]. Island annihilation was expected and confirmed by scanning tunneling microscopy (STM) [3] and in situ scanning electron microscopy (SEM) [4]. On the other hand, it is expected that monolayer-deep holes are formed by coalescence of the islands during growth which was confirmed by STM [5] and in situ SEM [6]. This paper reports their behavior during annealing after MBE growth.
* Corresponding author. Fax: +81 722 52 1163.
2. E x p e r i m e n t a l p r o c e d u r e
The experiment was done using an MBE-UHV SEM hybrid system [7]. Ga and As 4 were evaporated from Knudsen cells. A 25 keV electron beam was used for SEM observation and reflection high energy electron diffraction (RHEED) measurements. For in situ SEM imaging, the electron beam glancing angle was 10°. Thus the image was foreshortened in the vertical direction by 6 times. The (001) GaAs substrate was mounted by indium soldering on a silicon substrate. Sample heating was done by resistive heating of the silicon substrate. A smooth surface was prepared by buffer layer growth and annealing [4]. Observation of the growth and annealing processes was performed at about 580°C. The behavior of
0022-0248/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved SSDI 002 2-0248(95 )01064-5
N. Inoue et al. / Journal of Crystal Growth 164 (1996) 88-91
individual islands and holes was continuously observed by SEM with a resolution of about 5 nm. The observation rate was 10-80 s/frame. The image was recorded on a magneto-optic disk memory and a video tape.
3. Results An example of the growth and annealing processes is shown in Figs. 1 and 2. Fig. la shows the surface before growth. Several horizontal lines are
89
observed in the main part and irregular vertical lines are observed on the right side. These are monolayer steps. The electron beam was incident upward and collected by the detector on top of the image. The steps on the sunny side viewed from the detector appear dark. So, the central part was located at the bottom in this image. When the growth was started under the nucleation-and-growth mode monolayer islands appeared. In Fig. lb, islands appeared in four horizontal bands corresponding to the cycles indicated on the left side
1st layer 2nd layer 3rd layer 4th layer .......................
(a) Before growth
(c) 1'30" -~ 2'40"
.....
(b) Start of growth
(d) 2'50" -- end o f growth
Fig. 1. GaAs surface images before and during MBE growth reproduced from the video tape. The substrate temperature was about 580°C and the growth rate was 1 monolayer/18 s. The observation rate was 1 f r a m e / 8 0 s. (a) Before growth. (b) Start of growth. Growth cycles are shown on the left side. (c) During 1'30"-2'40". (d) 2'50"-end of growth.
N. Inoue et al. / Journal of C~stal Growth 164 (1996) 88-91
90
of the figure: the electron beam was scanned from top to bottom in 70 s. Because of the delay of nucleation, the surface was smooth in the first 1/3 of each growth cycle [8]. Islands (bright features of about 25 nm in size) and then holes (dark features of similar size with the islands) were observed. The surface got smoother again in the final stage of each cycle due to the shrinkage of holes. Steps are faintly observed because they were hindered by the islands and holes. As shown in Fig. 2c, bands are no longer visible in the next few cycles due to an unideal growth cycle. However, a few monolayer-deep holes were observed as marked by the white arrow showing that the cyclic 2D growth was still working. At the end of the growth of about 13 monolayers, a few multilayer-high islands were formed as noticed from a strong contrast marked by black arrows in Fig. 1d [4]. Immediately after the growth was terminated, islands and small holes were annihilated (not shown here but reported for islands in Ref. [4] and for holes in Ref. [9]). Fig. 2a shows the surface during 2'40"3'50" after growth termination. There were no islands and small holes. The size of the large holes was above 50 nm and is much larger than the average island spacing or hole size. This suggests that they were formed by the coalescence of small holes. In Fig. 2b the number of holes was apparently reduced compared to Fig. 2a. Thus, the smaller holes
(a) 2'40" ~ 3'50"
were shrinking. Coalescence of a hole to the adjacent hole was clearly observed as marked by the white arrow in the center of Fig. 2a. The other hole was connected to the steps as marked by the white arrow on the left side of Fig. 2a. Step retrieval and straightening took place simultaneously. It is to be noted that the big hole in the center marked by the black arrow in Fig. 2c grew in the horizontal ([110]) direction but shrank in the vertical ([1"10]) direction, as noticed from a comparison between the images in Figs. 2b and 2c. A similar anisotropic behavior occurred in the adjacent holes marked by the black arrows in Fig. 2b as noticed from a comparison with the images in Fig. 2a. In Fig. 2c, two holes were going to disappear as marked by the white arrows. The horizontal steps almost recovered their initial shape shown in Fig. la. Step straightening continued. In this case the observation was stopped here and the disappearance of the largest hole was not confirmed. But in the other case, it was observed that all the holes with the largest size of 0.2 microns were removed. It took 9 min to remove the largest hole.
4. Discussion It is shown that during annealing after MBE growth all the islands and most small holes disap-
(b) 4'00" - 5'10"
(e) 5'30" - 6'40"
i°l~j
Fig. 2. Surface images during annealing in an As4 ambient atmosphere after growth reproduced from the MO disk storage. (a) During 2'40"-3'50 after growth termination. (b) 4'00"-5' 10'2 (c) 5'30"-6'40".
N. Inoue et al. / Journal of Crystal Growth 164 (1996) 88-91
peared. There was still roughness, however, due to monolayer holes. And annihilation of large holes plays an important role in surface smoothing. The annealing time usually employed for growth interruption [2] was enough only for the annihilation of islands. But it was found that the time was not enough for the annihilation of holes. Because islands shrink only, but the holes grow and coalesce with each other. The anisotropic behavior of the holes is considered to be due to the tendency that atoms attach to As-terminated B-steps and detach from Gaterminated A-steps. In summary, the behavior of monolayer holes during annealing after growth was observed in situ by scanning electron microscopy. It was found that large holes were formed after annihilation of small islands. They grew and coalesce with each other on further annealing. They grow in the [110] direction but shrink in the [110] direction. Most holes finally shrink and disappear. Thus, the annihilation of holes plays an important role in the surface smoothing during annealing after growth.
91
Acknowledgements The authors are grateful to Ryuta Sakai for his help in sample preparation and Yoshihiro Imamura and Kosuke Ikeda for encouragement.
References [1] J.H. Neave, B.A. Joyce, P.J. Dobson and N. Norton, Appl. Phys. A 31 (1983) 1. [2] H. Sakaki, M. Tanaka and J. Yoshino, Jpn. J. Appl. Phys. 24 (1985) L417. [3] T. Ide, A. Yamashita and T. Mizutani, Phys. Rev. Lett. B46 (1992) 1905. [4] J. Osaka, Y. Homma and N. Inoue, J. Crystal Growth 150 (1995) 73. [5] J. Sudijono, M.D. Johnson, M.B. Elowitz, C.W. Snyder and B.G. Orr, Surf. Sci. 280 (1993) 247. [6] N. Inoue, J. Crystal Growth 146 (1995) 334. [7] Y. Homma, J. Osaka and N. Inoue, Jpn. J. Appl. Phys. 33 (1994) L563. [8] N. Inoue, J. Osaka and Y. Homma, J. Crystal Growth 150 (1995) 107. [9] N. Inoue, Y. Homma and J. Osaka, J. Vac. Sci. Technol., to be published.
ELSEVIER
Journal of Crystal Growth 164 (1996) 88-91
Behavior of monolayer holes on MBE grown GaAs surfaces during annealing revealed by in situ scanning electron microscopy N. Inoue
a, *
y . H o m m a b j. O s a k a c T. A r a k i a
a Research Institute for Advanced Science and Technology, Osaka Prefecture University, 1-2, Gakuencho, Sakai, Osaka 593, Japan b NTT Interdisciplinary Research Laboratories, 3-9-11, Musashino, Tokyo 180, Japan c NTT LSI Laboratories, 3-1, Morinosato-Wakamiya, Atsugi, Kanagawa 243-01, Japan
Abstract
The behavior of monolayer-deep holes during annealing after molecular beam epitaxy growth was observed in situ by scanning electron microscopy. Most of the small holes formed by coalescence of islands during growth shrink and disappear as soon as the islands disappear. However some holes are left unremoved and grow and coalesce with each other to form large holes. They grow in the [110] direction but shrink in the [170] direction. They finally shrink and disappear at about 10 min. It is shown that annihilation of holes plays an important role in surface smoothing during annealing after growth. The anisotropic behavior is attributed to the preference of detachment of Ga atoms from Ga-terminated A-steps and attachment to As-terminated B-steps.
1. I n t r o d u c t i o n
It is well known that annealing after MBE or CBE growth smoothes the surface [1]. This behavior was successfully utilized to obtain smooth interfaces in quantum wells [2]. Island annihilation was expected and confirmed by scanning tunneling microscopy (STM) [3] and in situ scanning electron microscopy (SEM) [4]. On the other hand, it is expected that monolayer-deep holes are formed by coalescence of the islands during growth which was confirmed by STM [5] and in situ SEM [6]. This paper reports their behavior during annealing after MBE growth.
* Corresponding author. Fax: +81 722 52 1163.
2. E x p e r i m e n t a l p r o c e d u r e
The experiment was done using an MBE-UHV SEM hybrid system [7]. Ga and As 4 were evaporated from Knudsen cells. A 25 keV electron beam was used for SEM observation and reflection high energy electron diffraction (RHEED) measurements. For in situ SEM imaging, the electron beam glancing angle was 10°. Thus the image was foreshortened in the vertical direction by 6 times. The (001) GaAs substrate was mounted by indium soldering on a silicon substrate. Sample heating was done by resistive heating of the silicon substrate. A smooth surface was prepared by buffer layer growth and annealing [4]. Observation of the growth and annealing processes was performed at about 580°C. The behavior of
0022-0248/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved SSDI 002 2-0248(95 )01064-5
N. Inoue et al. / Journal of Crystal Growth 164 (1996) 88-91
individual islands and holes was continuously observed by SEM with a resolution of about 5 nm. The observation rate was 10-80 s/frame. The image was recorded on a magneto-optic disk memory and a video tape.
3. Results An example of the growth and annealing processes is shown in Figs. 1 and 2. Fig. la shows the surface before growth. Several horizontal lines are
89
observed in the main part and irregular vertical lines are observed on the right side. These are monolayer steps. The electron beam was incident upward and collected by the detector on top of the image. The steps on the sunny side viewed from the detector appear dark. So, the central part was located at the bottom in this image. When the growth was started under the nucleation-and-growth mode monolayer islands appeared. In Fig. lb, islands appeared in four horizontal bands corresponding to the cycles indicated on the left side
1st layer 2nd layer 3rd layer 4th layer .......................
(a) Before growth
(c) 1'30" -~ 2'40"
.....
(b) Start of growth
(d) 2'50" -- end o f growth
Fig. 1. GaAs surface images before and during MBE growth reproduced from the video tape. The substrate temperature was about 580°C and the growth rate was 1 monolayer/18 s. The observation rate was 1 f r a m e / 8 0 s. (a) Before growth. (b) Start of growth. Growth cycles are shown on the left side. (c) During 1'30"-2'40". (d) 2'50"-end of growth.
N. Inoue et al. / Journal of C~stal Growth 164 (1996) 88-91
90
of the figure: the electron beam was scanned from top to bottom in 70 s. Because of the delay of nucleation, the surface was smooth in the first 1/3 of each growth cycle [8]. Islands (bright features of about 25 nm in size) and then holes (dark features of similar size with the islands) were observed. The surface got smoother again in the final stage of each cycle due to the shrinkage of holes. Steps are faintly observed because they were hindered by the islands and holes. As shown in Fig. 2c, bands are no longer visible in the next few cycles due to an unideal growth cycle. However, a few monolayer-deep holes were observed as marked by the white arrow showing that the cyclic 2D growth was still working. At the end of the growth of about 13 monolayers, a few multilayer-high islands were formed as noticed from a strong contrast marked by black arrows in Fig. 1d [4]. Immediately after the growth was terminated, islands and small holes were annihilated (not shown here but reported for islands in Ref. [4] and for holes in Ref. [9]). Fig. 2a shows the surface during 2'40"3'50" after growth termination. There were no islands and small holes. The size of the large holes was above 50 nm and is much larger than the average island spacing or hole size. This suggests that they were formed by the coalescence of small holes. In Fig. 2b the number of holes was apparently reduced compared to Fig. 2a. Thus, the smaller holes
(a) 2'40" ~ 3'50"
were shrinking. Coalescence of a hole to the adjacent hole was clearly observed as marked by the white arrow in the center of Fig. 2a. The other hole was connected to the steps as marked by the white arrow on the left side of Fig. 2a. Step retrieval and straightening took place simultaneously. It is to be noted that the big hole in the center marked by the black arrow in Fig. 2c grew in the horizontal ([110]) direction but shrank in the vertical ([1"10]) direction, as noticed from a comparison between the images in Figs. 2b and 2c. A similar anisotropic behavior occurred in the adjacent holes marked by the black arrows in Fig. 2b as noticed from a comparison with the images in Fig. 2a. In Fig. 2c, two holes were going to disappear as marked by the white arrows. The horizontal steps almost recovered their initial shape shown in Fig. la. Step straightening continued. In this case the observation was stopped here and the disappearance of the largest hole was not confirmed. But in the other case, it was observed that all the holes with the largest size of 0.2 microns were removed. It took 9 min to remove the largest hole.
4. Discussion It is shown that during annealing after MBE growth all the islands and most small holes disap-
(b) 4'00" - 5'10"
(e) 5'30" - 6'40"
i°l~j
Fig. 2. Surface images during annealing in an As4 ambient atmosphere after growth reproduced from the MO disk storage. (a) During 2'40"-3'50 after growth termination. (b) 4'00"-5' 10'2 (c) 5'30"-6'40".
N. Inoue et al. / Journal of Crystal Growth 164 (1996) 88-91
peared. There was still roughness, however, due to monolayer holes. And annihilation of large holes plays an important role in surface smoothing. The annealing time usually employed for growth interruption [2] was enough only for the annihilation of islands. But it was found that the time was not enough for the annihilation of holes. Because islands shrink only, but the holes grow and coalesce with each other. The anisotropic behavior of the holes is considered to be due to the tendency that atoms attach to As-terminated B-steps and detach from Gaterminated A-steps. In summary, the behavior of monolayer holes during annealing after growth was observed in situ by scanning electron microscopy. It was found that large holes were formed after annihilation of small islands. They grew and coalesce with each other on further annealing. They grow in the [110] direction but shrink in the [110] direction. Most holes finally shrink and disappear. Thus, the annihilation of holes plays an important role in the surface smoothing during annealing after growth.
91
Acknowledgements The authors are grateful to Ryuta Sakai for his help in sample preparation and Yoshihiro Imamura and Kosuke Ikeda for encouragement.
References [1] J.H. Neave, B.A. Joyce, P.J. Dobson and N. Norton, Appl. Phys. A 31 (1983) 1. [2] H. Sakaki, M. Tanaka and J. Yoshino, Jpn. J. Appl. Phys. 24 (1985) L417. [3] T. Ide, A. Yamashita and T. Mizutani, Phys. Rev. Lett. B46 (1992) 1905. [4] J. Osaka, Y. Homma and N. Inoue, J. Crystal Growth 150 (1995) 73. [5] J. Sudijono, M.D. Johnson, M.B. Elowitz, C.W. Snyder and B.G. Orr, Surf. Sci. 280 (1993) 247. [6] N. Inoue, J. Crystal Growth 146 (1995) 334. [7] Y. Homma, J. Osaka and N. Inoue, Jpn. J. Appl. Phys. 33 (1994) L563. [8] N. Inoue, J. Osaka and Y. Homma, J. Crystal Growth 150 (1995) 107. [9] N. Inoue, Y. Homma and J. Osaka, J. Vac. Sci. Technol., to be published.