Arsenic pressure dependence of hillock morphology on GaAs (n 1 1)A substrates grown using MBE

Arsenic pressure dependence of hillock morphology on GaAs (n 1 1)A substrates grown using MBE

Journal of Crystal Growth 227–228 (2001) 67–71 Arsenic pressure dependence of hillock morphology on GaAs (n 1 1)A substrates grown using MBE T. Ohach...

272KB Sizes 1 Downloads 15 Views

Journal of Crystal Growth 227–228 (2001) 67–71

Arsenic pressure dependence of hillock morphology on GaAs (n 1 1)A substrates grown using MBE T. Ohachia,*, M. Inadaa, K. Asaia, J.M. Fenga,b a

Department of Electrical Engineering, Doshisha University, Kyoto 610-0321, Japan b Computer Center of Gakushuin University, Tokyo 171-8588, Japan

Abstract Surface morphology of hillocks on the surface of various orientations of a GaAs substrate was observed by an atomic force microscope (AFM). Only (1 1 1)A and (2 1 1)A substrates formed pyramidal hillocks. The surfaces were the top layers of seven-period asymmetric double quantum wells or the buffer layer, and the Ga flux and substrate temperature were kept constant at 0.76 ML/s and 5208C, respectively. With increasing As pressure from 7.6 ML/s (beam equivalent pressure, BEP; 9.0  106 Torr) to 32 ML/s (BPE; 4.5  105 Torr), the hillock heights and slopes on (1 1 1)A first increased and then decreased at the highest pressures. The decrease in slope was large: at an As pressure of 1.95  105 Torr, the hillock slope on (1 1 1)A was about 0.88, whereas at 2.85  105 Torr, the slopes were usually less than 0.38. This reduction of the height and slope at higher As pressure is explained by the decreasing of the supersaturation of GaAs species in a quasi-liquid layer, QLL of As due to increment of Ga desorption. # 2001 Elsevier Science B.V. All rights reserved. Keywords: A1. Crystal morphology; A1. Desorption; A1. Growth models; A1. Quasi liquid layer; A3. Molecular beam epitaxy; B1. Gallium compounds

1. Introduction Demand for compound semiconductor devices grown by MBE and MOCVD has significantly increased because of their use in cellular phones and other devices, which are used in the field of information technology (IT). For high frequency and optical use, nanocrystals for self-organized quantum dots or wires on GaAs are especially important. Because the surfaces of these nanocrystals frequently contain orientations with high *Corresponding author. Tel.: +81-774-656329; fax: +81774-656811. E-mail address: [email protected] (T. Ohachi).

indices, the surface morphology of high-index GaAs surfaces should be better understood. Higher As pressures in MBE growth produce better quality epitaxial layers on high-index GaAs (n 1 1)A substrates [1], which improves the surface morphology of AlGaAs–GaAs asymmetric double quantum wells (ADQWs) grown on GaAs (n 1 1)A substrates [2]. The latter study showed that the heights of pyramidal hillocks on GaAs (1 1 1)A and (2 1 1)A substrate in AlGaAs–GaAs ADQWs decrease with increasing As pressure. How the As vapor pressure affects both the optical properties of quantum wells and the Si-doping behavior on GaAs (n 1 1)A substrates has also been studied [3–6].

0022-0248/01/$ - see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 0 1 ) 0 0 6 3 4 - 0

68

T. Ohachi et al. / Journal of Crystal Growth 227–228 (2001) 67–71

The sticking coefficient of Ga on GaAs surfaces is nearly unity if the substrate temperature is below 6508C. It was found that Ga desorption, which reduces the growth rate of GaAs, was suppressed at higher substrate temperatures by increasing As pressures or coexistence of Al [7–9]. However, using ADQWs and TEM observations [3,5], we found a different result: lower growth rates on (n 1 1)A, where n=1–4, at higher As pressures, which indicates increased Ga desorption rates. The PL peaks from ADQWs grown on the (4 1 1)A surface were blue-shifted or shifted towards higher energies due to thinning of the smaller quantum wells [5]. The orientation dependence of Ga desorption varies from (1 0 0), which is about 4.9% of the incident Ga flux, to that of (1 1 1)A, which is about 4.4% [6]. The growth rate reduction on As pressure dependence of Ga desorption was explained by the possibility of a quasi-liquid layer (QLL) of As [5,6]. This is consistent with the amorphous As features on high-coverage GaAs (0 0 1)cð4  4Þ as suggested by Larsen et al. [10]. In this study, growth hillocks on GaAs (1 1 1)A and GaAs (2 1 1)A buffer layers over AlGaAs– GaAs ADQWs were measured ex situ with an AFM as a function of As pressure. We also discuss how a QLL of As might explain these results.

2. Experiment procedure The samples were fabricated by MBE in a VG Semicon V80H chamber on semi-insulating GaAs (1 1 1)A, (2 1 1)A, and (1 0 0) substrates. Prior to loading, the substrates were cleaned using Semicocleaning etchant and deionized water. After outgassing at 4508C for 1 h in a preparation chamber, the substrates were transferred to the growth chamber where the thin oxide film was desorbed by heating at 6808C for 5 min under As4 flux bombardment. The substrates were then cooled to 5208C and a 1 mm undoped GaAs buffer layer was grown at 0.76 ML/s. This is equivalent to a Ga flux of 4.61  1014 atoms/(cm2 s). As in the previous study [2], the structure of our ADQWs was 7 periods of the following sequence: an undoped 14-nm wide well, a 5-nm Al0.35Ga0.65As tunnel-barrier, an 8-nm narrow

well, with a 20-nm layer of Al0.35Ga0.65As. To see if Al influenced the hillock heights, we also grew a 1-mm-thick buffer layer of GaAs over the ADQWs. During the MBE growth of all samples, the Ga, Al fluxes, and substrate temperature were kept constant. Assuming a monolayer thickness of 0.5653 nm, a growth rate of 1 ML/s of GaAs (1 0 0) equals a Ga flux of 6.26  1014 atoms/(cm2 s) and a Ga BEP of 1.4  106 Torr. We used RHEED oscillations on (1 0 0) to calibrate the Ga flux as a function of Ga K-cell temperature. Conversely, the As flux of 4.6  1015 and 19.7  1015 atoms/(cm2 s) corresponds to BEPs of 9.0  106 and 4.5  105 Torr, respectively. The AFM measurements were done with a Shimazu model SPM9500.

3. Results and discussion The surface morphology of the samples on the different substrates depended strongly on the As pressure. Fig. 1 shows the height change of hillocks on (1 1 1)A and (2 1 1)A surfaces obtained from the previous measurements of ADQWs surfaces [2]. Pyramidal hillocks formed on only (1 1 1)A and (2 1 1)A substrates, whereas (1 0 0), (3 1 1)A, and (4 1 1)A had rounded hillock features or striations [2]. On (1 1 1)A, as the As pressure increased from 1.95  105 Torr, the hillock height first remained nearly constant and then decreased

Fig. 1. The maximum height between the lowest and highest points in a typical 5 mm  5 mm section of surface as a function of As pressure. The surfaces are GaAs (1 1 1)A and (2 1 1)A surfaces in an ADQW. The data is from Ref. [2].

T. Ohachi et al. / Journal of Crystal Growth 227–228 (2001) 67–71

69

Fig. 2. AFM images of triangle pyramids and several cross-sectional profiles for (1 1 1)A surfaces on a GaAs buffer layer over ADQWs that were similar to those represented in Fig. 1: (a) growth with As pressure of 1.95  105 Torr and (b) growth with As pressure equal to 2.85  105 Torr. The end points (A-F) of the 2D profile plots are shown in the AFM plan views. Adjacent to each surface and plan view are plots of the fraction of surface at each height.

at the highest As pressures. On (2 1 1)A, the height monotonically decreased in this pressure range. The (1 1 1)A surface has pyramidal hillocks in which three side walls have equivalent crystallographic faces near (1 1 1)A. At higher As pressure, the pyramidal pattern completely disappeared and the surfaces became very flat. Rarely, we observed extremely high, four-sided hillocks on (1 1 1)A at an As pressure of 1.5  106 Torr. We do not know the source of either type of hillock. The AFM measurements on the buffer layer agree with the data in Fig. 1. Figs. 2a and b show typical pyramidal hillocks on a (1 1 1)A surface of

the buffer layers grown at As pressures of 1.95  105 and 2.85  105 Torr, respectively. The height change of pyramidal hillocks is consistent with the result in Fig. 1. The slope of the pyramids is less than 18, which corresponds to a large step–step distance on (1 1 1)A. Although we cannot explain this large decrease in hillock height and hillock slope with increasing As pressures, a large As surface coverage as QLL could have a significant effect on several growth processes: the transport of Ga though the QLL to the growing surface, the step-edge energy, Ga incorporation at steps, and surface diffusion of Ga on terraces and steps.

70

T. Ohachi et al. / Journal of Crystal Growth 227–228 (2001) 67–71

Fig. 2. (Continued).

The increase of Ga desorption from the surface at higher As pressures arose from an increase of coverage in the form of a QLL [5,6]. When the As pressure increases or the substrate temperature decreases, the surface reconstruction of GaAs (1 0 0) changes from (2  4) to cð4  4Þ and gradually increases its As coverage to 1.75 [11,12]. The As pressures used here are higher than those under which the amorphous coverage of 7 ML [10] is reported a chemisorbed structure on cð4  4Þ. Thus, a surface coverage of 1.75 or greater is expected. The order of 1.0  105 Torr, which is a much higher vapor pressure of disappearing RHEED patterns is large enough to form large coverage of As over the GaAs (0 0 1)cð4  4Þ surface. The surface coverage

state of As is affected by the amount of Ga flux, but the RHEED pattern could detect only the order of the GaAs surface. As our previous measurements of ADQW well-thickness indicated that the GaAs growth rates on (1 1 1)A and (2 1 1)A decreased by only about 4% at the highest As pressures used here, the most likely explanation for the reduction in hillock slope might involve the change of Ga chemical potential or supersaturation of GaAs species on the growing surface within the QLL. At higher As pressures, it is possible to form a QLL of As on the surface [5,6]. The thickness of this QLL probably increases with the As pressure and the Ga desorption should be greater for a thicker QLL because a thicker QLL of As would keep the Ga atoms away from the

T. Ohachi et al. / Journal of Crystal Growth 227–228 (2001) 67–71

GaAs interface. Therefore, Ga desorption rates should increase at high As pressures and low temperatures. The increase of Ga desorption from the surface at high As pressures arose from an increasing coverage with a QLL. The slope of (1 1 1)A pyramidal surface was less than 18 and decreased at higher As pressure. This reduction of the slope angle at higher As pressure was explained by the decreasing of the supersaturation of GaAs species in QLL, that the steps on the hillocks had equivalent crystallographic directions on average suggests that the step-edge energies on (1 1 1)A and (2 1 1)A depend on orientation.

71

area of Atomic-Scale Surface and Interface Dynamics (JSPS-RFTF97P00201). We were also supported by the project, ‘‘Nano Structure Hybrid Devices and their Properties’’ at the Research Center for Advanced Science and Technology (RCAST) of Doshisha University. We also acknowledge the support of special research grants for the development of characteristic education from the promotion and mutual aid corporation for private schools, Japan. One of the authors (T.O.) acknowledges support from Dr. J.T. Nelson for valuable comments and from the Foundation for the Promotion of Material Science and Technology of Japan (MST).

4. Conclusions AFM results show that the surface morphology of GaAs substrates depends strongly on the As pressure. The (1 1 1)A surface has pyramidal hillocks in which three sidewalls have equivalent crystallographic faces of slope less than 18. When the As pressure increased, the height of the pyramids decreased. In this range of As pressures, the As coverage could be greater than 1.75, and hence several As-surface-coverage effects are possible. One explanation is that the GaAs supersaturation in a QLL on growing GaAs surface decreases with the QLL thickness, which increases at higher As pressures. An increase of Ga desorption from the surface at high As pressures can also arise from an increasing coverage with an As QLL, but a satisfactory explanation of the results here remain for a future study.

Acknowledgements This work was supported by the Japan Society for the Promotion of Science (JSPS) Fund of Research for the Future Program (RFTF) in the

References [1] K. Fujita, P.O. Vaccaro, Curr. Topics Crystal Growth Res. 3 (1997) 1. [2] T. Ohachi, J.M. Feng, K. Asai, M. Uwani, M. Tateuchi, P.O. Vaccaro, K. Fujita, Microelectron. J. 30 (1999) 471. [3] T. Ohachi, J.M. Feng, K. Asai, M. Uwani, M. Tateuchi, P.O. Vaccaro, K. Fujita, J. Crystal Growth 201/202 (1999) 226. [4] J.M. Feng, K. Asai, Y. Narukawa, Y. Kawakami, S. Fujita, T. Ohachi, Appl. Surf. Sci. 159/160 (2000) 532. [5] T. Ohachi, J.M. Feng, K. Asai, J. Crystal Growth 211 (2000) 405. [6] K. Asai, J.M. Feng, P.O. Vaccaro, K. Fujita, T. Ohachi, Appl. Surface Sci. 159/160 (2000) 301. [7] E.M. Gibson, C.T. Foxon, J. Zhang, B.A. Joyce, Appl. Phys. Lett. 57 (1990) 1203. [8] T. Suzuki, I. Ichimura, T. Nishinaga, Jpn. J. Appl. Phys. 30 (1991) L1612. [9] R. Fischer, J. Klem, T.J. Drummond, R.E. Thorne, W. Kopp, H. Morkoc, A.Y. Cho, J. Appl. Phys. 54 (1983) 2508. [10] P.K. Larsen, J.H. Neave, J.F. van der Veen, P.J. Dobson, B.A. Joyce, Phys. Rev. 27 (1983) 4966. [11] T. Hashizume, Q.K. Xue, J. Zhou, A. Ichimiya, T. Sakurai, Phys. Rev. Lett. 72 (1994) 2208. [12] V.V. Preobrazhenskii, M.A. Putyato, O.P. Pchelyakov, B.R. Semyagin, J. Crystal Growth 201/202 (1999) 170.