Structural properties of GaN layers on Si(001) grown ...

2 downloads 52 Views 2MB Size Report
Paul-Drude-Institut für Festkörperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany. Received 28 October 1997; accepted for publication 5 January ...
Structural properties of GaN layers on Si(001) grown by plasma-assisted molecular beam epitaxy B. Yang, A. Trampert, O. Brandt, B. Jenichen, and K. H. Ploog Citation: J. Appl. Phys. 83, 3800 (1998); doi: 10.1063/1.367144 View online: http://dx.doi.org/10.1063/1.367144 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v83/i7 Published by the American Institute of Physics.

Related Articles Non-adiabatic ab initio molecular dynamics of supersonic beam epitaxy of silicon carbide at room temperature J. Chem. Phys. 138, 044701 (2013) LaAlO3/Si capacitors: Comparison of different molecular beam deposition conditions and their impact on electrical properties J. Appl. Phys. 113, 034106 (2013) N incorporation in GaInNSb alloys and lattice matching to GaSb J. Appl. Phys. 113, 033502 (2013) Properties of epitaxial BaTiO3 deposited on GaAs Appl. Phys. Lett. 102, 012907 (2013) Step-step interactions on GaAs (110) nanopatterns J. Appl. Phys. 113, 024309 (2013)

Additional information on J. Appl. Phys. Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

JOURNAL OF APPLIED PHYSICS

VOLUME 83, NUMBER 7

1 APRIL 1998

Structural properties of GaN layers on Si„001… grown by plasma-assisted molecular beam epitaxy B. Yang,a) A. Trampert, O. Brandt, B. Jenichen, and K. H. Ploog Paul-Drude-Institut fu¨r Festko¨rperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany

~Received 28 October 1997; accepted for publication 5 January 1998! We report on the growth and microstructure of GaN films deposited on Si~001! substrates by plasma-assisted molecular beam epitaxy. GaN films grown directly on Si~001! are found to be phase mixtured, containing both cubic ~b! and hexagonal ~a! modifications. The origin of this phase mixture is identified to be due to the formation of amorphous Six Ny at the GaN/Si interface during the nucleation stage. Therefore, a GaAs buffer layer is employed to prevent the formation of Six Ny . GaN films grown on this GaAs/Si~001! structure are in fact predominantly cubic and exhibit the characteristic band-edge photoluminescence ~PL! of b-GaN up to room temperature. However, the PL efficiency from these samples is low compared to that of b-GaN layers directly grown on GaAs~001!. We explain the lower PL efficiency by the presence of additional structural defects, which are observed in the GaN/GaAs/Si~001! heterostructure by transmission electron microscopy. © 1998 American Institute of Physics. @S0021-8979~98!05907-6#

GaAs/Si~001! and the observation of the b-GaN band-edge photoluminescence from these samples.

I. INTRODUCTION

For almost a decade, material scientists have attempted to grow high-quality GaN layers on Si, motivated by the irreplaceable advantages offered by Si over other substrates, such as low price, high crystal and surface quality, largearea-wafer availability, well-established preparation and processing, high thermal stability and conductivity, and so on.1–3 However, growth of GaN directly on Si~111! frequently resulted in phase-mixtured films.4 Single-phase hexagonal ~a! GaN films have only been realized if buffer layers, such as AlN ~Ref. 5! or GaAs ~Ref. 6! were used. On the other hand, growth of GaN directly on Si~001! has been found to generally result in phase-mixtured7,8 or even polycrystalline films.3 The actual reasons for these disappointing findings are not yet well understood. It is presumed that the large lattice mismatch ~>17%! between GaN and Si, and/or the common problems in the epitaxy of polar materials on nonpolar substrates, and possibly together with the poor wetting of GaN on Si, play important roles here.7 In addition, due to the metastable nature of cubic ~b! GaN, the phase mixture is likely to occur particularly for the case of GaN on Si~001!, as already reported in previous studies.7,8 In order to develop procedures for improving the quality of the GaN films, it is obviously necessary to study both the basic growth mechanisms of GaN on Si and the origin of the phase mixture. Here, we study the direct nucleation of b-GaN on Si~001! and investigate the origin of the phase mixture. We find that amorphous Six Ny formed at the GaN/Si interface during the nucleation stage is responsible for the phase mixture. Therefore, we employ a buffer layer to prevent the formation of Six Ny at the GaN/Si interface. The effect of the GaAs buffer layer used in our experiments is highlighted by the improved phase purity of the b-GaN films grown on

II. EXPERIMENT

The samples studied in this paper are grown in a homemade molecular beam epitaxy ~MBE! system equipped with a dc glow discharge N plasma source operating with a power of 30 W.9,10 Two series of p-type Si~001! wafers with miscuts of 1.7° and 4° toward ^110& are used as substrates. Prior to loading into the MBE chamber, the substrates are degreased in solvents and etched in a buffered HF solution. In the growth chamber, the Si substrates are annealed at 850 °C for half an hour prior to growth. Then, GaN deposition is started by opening both the Ga and N cell shutters and starting the N plasma simultaneously. The substrate temperature and the growth rate are 650 °C and 0.06 ML/s, respectively. In situ reflection high-energy electron diffraction ~RHEED! and ex situ double-crystal x-ray diffractometry ~DCXRD!, transmission electron microscopy ~TEM!, and atomic force microscopy ~AFM! are used to investigate the crystal structure and morphology of the GaN films. The x-ray diffractometer is equipped with a CuK a anode and a Ge~004! monochromator. X-ray measurements are taken either with a wide open detector @x-ray rocking curve ~XRC!# or with two 0.5 mm detector slits ~v or v-2u scan, angular acceptance ;0.1°!. The optical properties of the GaN epilayers are investigated by means of photoluminescence ~PL! measurements using the 325 nm line of a He–Cd laser as excitation source at temperatures ranging from 4.2 to 300 K. The TEM cross-sectional specimens are prepared from the as-deposited GaN layers by mechanical prethinning followed by Ar1 ion milling using a specimen cooling unit. Both conventional and high-resolution TEM observations are carried out in Jeol JEM 4000FX and EX microscopes operating at 400 kV accelerating voltage.

a!

Electronic mail: [email protected]

0021-8979/98/83(7)/3800/7/$15.00

3800

© 1998 American Institute of Physics

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

J. Appl. Phys., Vol. 83, No. 7, 1 April 1998

Yang et al.

3801

FIG. 1. In situ RHEED patterns of a 10 ML GaN nucleation layer grown at 500 °C with an effective N/Ga flux ratio of 2.0 taken along both @110# and @100# azimuths.

III. RESULTS AND DISCUSSION A. GaN films grown directly on Si„001…

We first discuss the direct growth of epitaxial b-GaN layers on Si~001! substrates. The GaN films are deposited in two steps, which include a relatively low-temperature ~500– 550 °C! nucleation layer followed by high-temperature growth ~650 °C! for the rest of the film. Based on our experience of growing b-GaN on GaAs~001!,9,10 we focus on studying growth conditions under which a thin ~10 ML! epitaxial b-GaN layer forms and preferably completely covers the Si~001! surface. Such an epitaxial ‘‘wetting’’ layer is expected to not only favor a single-phase b-GaN layer, but also to result in a smooth and homogeneous surface morphology,7 i.e., it is a prerequisite for the growth of highquality GaN layers. However, such a ‘‘wetting’’ layer is much more difficult to realize than in the case of b-GaN layers grown on GaAs. No epitaxial nucleation layer at all is obtained in our experiments on 1.7° misoriented Si~001! substrates, which display clear double-domain ~232! RHEED patterns after annealing.11 The surface morphology of these nucleation layers consists of isolated islands. This islanding of GaN on the 1.7° misoriented Si~001! surface is ascribed to the formation of antiphase domains which might inhibit the coalescence process of the initial GaN nuclei, thus resulting in substrate nitridation, which consequently degrades the substrate surface. In the following we, therefore, exclusively use 4° misoriented Si~001! wafers as substrates for the growth of the GaN samples. These substrates exhibit a single-domain ~132! reconstruction after annealing,11 and we have in fact succeeded in obtaining a crystalline GaN nucleation layer in a narrow range of growth conditions, namely, at a substrate temperature of about 500–550 °C and an effective N/Ga flux ratio of 2.0. The RHEED patterns shown in Fig. 1, which are taken along both @110# and @100# azimuths, demonstrate the cubic crystal structure of a 10 ML GaN nucleation layer deposited under these conditions. The AFM image of this sample shown in Fig. 2 reveals a smooth surface morphology with a peak-to-valley roughness of only 0.8 nm, suggesting the thin GaN layer is smooth and connected. Despite this epitaxial b-GaN nucleation layer, subsequent growth of GaN results in a clear transition from the b phase to a phase mixture of both a and b phases as revealed by RHEED. This transition is found to be independent of both the thickness of the GaN nucleation layer and the growth conditions used for the subsequent growth stage.

FIG. 2. AFM image of the 10 ML GaN nucleation layer. The peak-to-valley roughness is 0.8 nm.

Upon further growth, the a phase becomes eventually dominant. As expected from these observations, the x-ray v –2u scan ~Fig. 3! of a 800 nm thick GaN film is dominated by the a-GaN~0002! reflection. The b-GaN~002! reflection is not detectable at the expected angular position in the spectrum, instead, two peaks are present whose origins are not yet clarified. The v scan of the a-GaN~0002! reflection reveals that its full width at half maximum amounts to 2°–3°, demonstrating the large spread in the out-of-plane orientation between the individual a-GaN domains in the film ~see the inset of Fig. 3!. Azimuthal scans across asymmetric reflections show the in-plane orientation of the a-GaN columns to be close to random. In order to study the origins of both the formation of a-GaN and the observed phase transformation, TEM investigations of the microstructure of the GaN/Si interface are

FIG. 3. X-ray v –2u scan of a 800 nm thick GaN film grown directly on Si~001!. The inset shows the v scan spectrum of the a-GaN~0002! reflection.

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

3802

Yang et al.

J. Appl. Phys., Vol. 83, No. 7, 1 April 1998

plane orientation spread of the columns is less than 4° ~in agreement with the x-ray measurements!, resulting in a highly textured, fiber-like microstructure. The change of contrast close to the GaN/Si interface originates from grains of b-GaN, as demonstrated by means of the high-resolution TEM image ~see the inset of Fig. 4! and selected-area electron diffraction ~SAD! patterns ~not shown here!. These small b-GaN domains are not fully connected with each other as to form a continuous b-GaN layer, but are intersected by randomly distributed and even smaller a-GaN domains. With increasing film thickness, the a-GaN domains extend laterally, and finally overgrow the b-GaN domains, resulting in the phase transformation as observed by in situ RHEED. In the following, we will discuss in detail the origin of the phase mixture and the observed phase transformation in these films. B. Origin of the phase mixture

FIG. 4. Cross-sectional TEM dark-field image of the 800 nm thick GaN film grown directly on Si~001!. The inset displays a high-resolution image of a small b-GaN domain at the interface.

performed. The cross-sectional TEM dark-field micrograph in Fig. 4 shows the microstructure of the 800 nm thick GaN layer characterized by the x-ray in Fig. 3. Close to the surface, the TEM image displays the well-known columnar growth morphology of a-GaN. The columns are separated by straight boundaries and have well-defined flat surface facets. Within these a-GaN grains, a high density of planar defects is detected, which are identified to be stacking faults lying in the close-packed basal plane. These stacking faults are running almost parallel to the GaN/Si interface, and the out-of-

In order to ascertain the origin of the phase mixture, phase transformation and the textured structure of the GaN film, the microstructure of the GaN nucleation layer, together with the GaN/Si interface are studied in detail. Figure 5 shows a cross-sectional view of the GaN nucleation layer whose structure and morphology have already been characterized by RHEED and AFM ~cf. Figs. 1 and 2!. The largescale overview of the heterostructure in Fig. 5~a! demonstrates that the 10 ML GaN deposit indeed forms a quasiconnected layer. Up- and downward arrows indicate regions directly at the interface and within the layer, respectively, which exhibit a contrast different from that of b-GaN. Some portions of this micrograph are shown in higher magnification in Figs. 5~b!, 5~c!, and 5~d! and reveal the following interesting phenomena. The nucleation layer is actually not a homogenous and continuous GaN layer, but is inter-

FIG. 5. Cross-sectional TEM images of the 10 ML thick GaN nucleation layer: ~a! overview, upward, and downward arrows are explained in the text; ~b! higher magnification image of the region where bright contrast is observed both between GaN nuclei and at the GaN/Si interface; ~c! higher magnification image showing a a-GaN domain nucleating on top of an amorphous patch; and ~d! high-resolution image showing a part of the interface where good epitaxial relation between b-GaN and Si is achieved, the arrows indicate the misfit dislocations formed every five Si lattice planes.

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

J. Appl. Phys., Vol. 83, No. 7, 1 April 1998

rupted by randomly distributed amorphous patches @as indicated by the downward arrow in Fig. 5~b!# inbetween the b-GaN grains. At some GaN/Si interfacial regions @indicated by the upward arrow in Fig. 5~b!#, the b-GaN grains are heavily misoriented and contain a very high density of stacking faults. a-GaN grains whose basal plane is parallel to the Si~001! surface are observed to grow on top of the amorphous patches at the interface @Fig. 1~c!#. For those GaN/Si interfacial regions where a good epitaxial relation between GaN and Si is established @Fig. 1~d!#, the large lattice mismatch ~>17%! between b-GaN and Si is accommodated by misfit dislocations, which form a 5:6 coincidence lattice as it might be expected from the lattice constant ratio. Concluding from the above TEM images shown in Figs. 4 and 5, there are mainly two sources for a-GaN formation in the film. First, on top of the amorphous material where the connection between GaN and Si is lost, a-GaN islands nucleate preferentially. Second, in those interfacial regions where GaN and Si are not in perfect registry with each other, misoriented GaN grains together with a high density of stacking faults can also induce the nucleation of a-GaN, as we have found in the growth of b-GaN on GaAs~001!.9,10 Finally, it is clear that once the phase mixture occurs, the initial a-GaN domains extend laterally upon further growth, and eventually overgrow the b-GaN domains, perhaps due to the higher growth rate of the basal plane together with the lower surface and total energy, as compared with the ~001! face of b-GaN. As most of the a-GaN domains nucleate on top of the amorphous material, it is understandable that the in-plane orientation of the a-GaN grains is random. The next question we have to answer is the composition and origin of the amorphous material in the GaN/Si interfacial region. In fact, a previous TEM study of GaN grown on Si~001! also displayed an amorphous-like layer at the GaN/Si interface together with small a-GaN domains.12 However, no attention was paid to the interfacial layer. Because of the high chemical reactivity between Si and atomic N, we assume that the formation of amorphous Six Ny during the nucleation stage is likely to occur.13 In the following, we will give experimental evidence for this assumption. Figure 6 shows two cross-sectional high-resolution TEM images, where on the one hand, a GaAs layer is grown directly on Si~001! @Fig. 6~a!#, and on the other hand, the Si~001! substrate was first nitrided by a N plasma, followed by the deposition of a 100 nm thick GaAs cap layer @Fig. 6~b!#. The dose of the N plasma used for the nitridation of the Si surface is equivalent to that used for the growth of a 10 ML GaN buffer layer. It is clear that a good epitaxial relation is achieved in the GaAs/Si~001! system in Fig. 6~a!. In contrast, in Fig. 6~b!, a homogeneous amorphous layer is found to be sandwiched between the polycrystalline GaAs layer and the Si substrate. As the chemical and thermal treatment for both substrates were the same, we conclude that the amorphous material has to be amorphous Six Ny . Other evidence which favors our conclusion comes from RHEED. In both cases, after annealing, clear and streaky ~132! RHEED patterns are observed, demonstrating that the Si~001! surfaces are clean and well ordered before nitridation. However, once the nitridation process starts by supplying N to the

Yang et al.

3803

FIG. 6. ^110& cross-sectional high-resolution TEM images of ~a! the GaAs layer grown on Si~001! and ~b! the GaAs layer grown on a nitrided Si~001! substrate.

Si~001! surface @for the sample shown in Fig. 6~b!#, the intensity of the RHEED patterns decreases and finally vanishes, indicating the formation of an amorphous layer covering the Si~001! surface. Therefore, we conclude that the initial growth of b-GaN on Si~001! by plasma-assisted MBE undergoes two major processes: the nucleation of threedimensional b-GaN nuclei on Si~001! and, simultaneously, the formation of amorphous Six Ny . That is to say, both the amorphous Six Ny material and the b-GaN nuclei are expected to coexist in the GaN nucleation layer. As a consequence, the phase mixture is inevitable unless proper buffer layers are used to prevent the Six Ny formation at the nucleation stage. C. GaN layers grown on GaAs-buffered Si„001…

In order to inhibit the formation of Six Ny at the nucleation stage, a GaAs buffer layer is employed in our experiments. Optimized growth conditions are used to grow both the GaAs buffer layer and the b-GaN epilayers.14,15 The xray v –2u profile of a 1 mm thick GaN film grown on GaAsbuffered Si~001!, shown in Fig. 7, evidences that this layer is in fact of cubic structure. No a-GaN reflection is observed.

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

3804

Yang et al.

J. Appl. Phys., Vol. 83, No. 7, 1 April 1998

FIG. 7. X-ray v –2u scan of a 1 mm thick GaN film grown on GaAs/Si~001!. The inset shows a selected-area electron diffraction pattern taken from a ^110& cross-sectional sample of the GaN/GaAs heterostructure.

The half width of the b-GaN~002! reflection is about 15 arcmin, which is about three times as large as that of the b-GaN layers grown directly on GaAs~001! substrates.16 A similar quantitative relation is found for XRC scans, demonstrating that both the mosaicity and the inhomogeneous strain are increased by the use of Si~001! as substrate for b-GaN growth. The phase content and crystallinity of the GaN film are also characterized by taking SAD patterns from a ^110& cross-sectional sample of the GaN/GaAs heterostructure ~see the inset of Fig. 7!. The superposition and the perfect alignment of the diffraction patterns of the GaAs buffer layer and the GaN film along their ^110& zone axes visualizes the epitaxial orientation relationship. The lattice mismatch of 17% measured from the distance of the corresponding GaN and GaAs diffraction spots is in agreement with the result obtained by DCXRD. The streaks along the ^111& directions indicate that the major structural defects in the film are stacking faults. These results of our structural analysis are essentially identical to those obtained for b-GaN layers grown directly on GaAs~001!.17 The surface morphology of this 1 mm thick GaN layer grown on the GaAs-buffered Si~001! substrate is shown in Fig. 8. The characteristic brick-shaped features are similar to the samples observed for b-GaN grown directly on GaAs~001!. The peak-to-valley roughness amounts to 20 nm, which is larger compared to the GaN films of the same thickness grown directly on GaAs~001!. Figure 9 displays the PL spectra of the 1 mm thick GaN layer. At temperatures below 100 K, a peak centered at about 3.16 eV is dominant. The temperature dependence of this peak demonstrates that it originates from a donor–acceptor pair transition. When the temperature is increased to 100 K, the band-edge emission of b-GaN ~Ref. 18! takes over. However, with further increase of temperature to 300 K, a broad peak in the green spectral region ~centered at 2.5 eV! becomes dominant. Furthermore, the integrated PL intensity of this sample is only one-tenth of that of comparable GaN layers grown directly on the GaAs~001! substrate. The presence of the green luminescence band as well as the low PL intensity suggests that the use of a Si~001! substrate, even

FIG. 8. AFM image of the 1 mm thick GaN film grown on GaAs/Si~001!.

with an additional GaAs buffer layer, introduces additional defects, which act as deep centers creating additional radiative and nonrecombination channels in the GaN layer. Obviously, the identification of these defects is important for the development of procedures to reduce their density or even to inhibit their formation. Actually, TEM reveals two additional sources of structural defects in the GaN film discussed above as compared

FIG. 9. PL spectra of the 1 mm thick GaN film grown on GaAs/Si~001! measured at various temperatures.

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Yang et al.

J. Appl. Phys., Vol. 83, No. 7, 1 April 1998

FIG. 10. Cross-sectional high-resolution TEM images taken along the ^110& direction: ~a! GaAs/Si interface. ~b! GaN/GaAs interface.

with that of the GaN epilayers grown directly on GaAs~001!.16,17 Figures 10 and 11 display cross-sectional TEM micrographs of the b-GaN/GaAs/Si heterostructure. The high-resolution images @Figs. 10~a! and 10~b!# taken from both the GaAs/Si and GaN/GaAs interfaces show wellestablished epitaxial relationships. The large lattice mismatches at both interfaces are primarily accommodated by misfit dislocations. Furthermore, stacking faults and microtwins penetrate into the epilayers. These findings are similar to previously reported results.16 However, the TEM micrograph shown in Fig. 11~a!, which covers an interfacial length of 5.2 mm reveals the presence of bundles of threading dislocations originating at the GaAs/Si interface and running through the GaAs layer into the GaN epilayer. These dislocations are of 60° type, and are not present in b-GaN grown directly on GaAs~001!.16 We believe these defects to result from the thermal misfit within the three-compound system ~GaN/ GaAs/Si!. Occasionally, even cracks are observed, which cut through the GaN/GaAs layers down to the GaAs/Si interface.

3805

FIG. 11. Cross-sectional TEM images of the GaN/GaAs/Si~001! heterostructure taken along the ^110& direction: ~a! 5.2 mm scale GaN/GaAs/ Si~001! interface overview. The arrows indicate bundles of 60° dislocations originating at the GaAs/Si interface. ~b! 1.3 mm scale GaN/GaAs interface. The arrows indicate the threading dislocations introduced into the GaN layer at the GaN/GaAs interface.

Threading dislocations originating from the GaN/GaAs interface and operating on a nm scale are visible in Fig. 11~b!. The density of these threading dislocations does not decrease with thickness. We speculate that the origin of these threading dislocations, which are not observed for b-GaN layers on on-axis GaAs~001! substrates, is related to the vertical mismatch between GaN and GaAs at the steps on the off-axis GaAs surface. The origin of this type of defect is under further investigation in our laboratory. IV. SUMMARY

In conclusion, we have demonstrated that though the near-coincidence lattice model is practicable to the epitaxy of b-GaN on Si~001! ~lattice mismatch>17%, a Si :a GaN >6:5!, the nucleation of b-GaN directly on Si~001! is obstacled by the formation of amorphous Six Ny in the GaN/Si interfacial region. The amorphous Six Ny in this region leads to a phase-mixtured and a textured microstructure of GaN films grown directly on Si~001!. Therefore, buffer layers

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

3806

Yang et al.

J. Appl. Phys., Vol. 83, No. 7, 1 April 1998

such as GaAs, SiC, etc., are considered to be necessary to improve the quality of the GaN films grown on Si~001!. The GaAs buffer layer employed in our experiment is demonstrated to be useful for the growth of b-GaN epilayers on Si~001! substrates. However, threading dislocations of quite high density, which are induced by the use of the Si~001! substrate, exist in the GaN layer. These defects seriously degrade the structural and optical quality of the b-GaN layers. Obviously, further optimization of the growth conditions, or employment of other kinds of buffer layers, such as SiC, are necessary to get high-quality b-GaN layers on Si~001! substrates.

ACKNOWLEDGMENTS

The authors would like to thank H. P. Scho¨nherr and P. Schu¨tzendu¨be for their valuable technical assistance with the setup of our MBE system, U. Jahn and J. Menniger for the scanning electron microscope measurement, and M. Wassermeier for the help with the AFM measurements. The authors greatly acknowledge the Forschungszentrum Ju¨lich, Institut fu¨r Mikrostrukturforschung, for providing their microscope facilities. This work was partially supported by the Bundesministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie of the Federal Republic of Germany.

1

T. Lei, M. Fanciulli, R. J. Molnar, T. D. Moustakas, R. J. Graham, and J. Scanlon, Appl. Phys. Lett. 59, 944 ~1991!. 2 T. Lei and T. D. Moustakas, Mater. Res. Soc. Symp. Proc. 242, 433 ~1992!. 3 Z. Sitar, M. J. Paiseley, B. Yan, and R. F. Davis, Mater. Res. Soc. Symp. Proc. 162, 537 ~1990!. 4 T. D. Moustakas, T. Lei and R. J. Molnar, Physica B 185, 36 ~1993!. 5 M. Godlewski, J. P. Bergerman, B. Monemar, U. Rossner, and A. Barski, Appl. Phys. Lett. 69, 2089 ~1996!. 6 J. W. Yang, C. J. Sun, Q. Chen, M. Z. Anwar, M. Asif Khan, S. A. Nikishin, G. A. Seryogin, A. V. Osinsky, L. Chernyak, H. Temkin, C. Hu, and S. Mahajan, Appl. Phys. Lett. 69, 3566 ~1996!. 7 T. Lei, T. D. Moustakas, R. J. Graham, Y. He, and S. J. Berkowitz, J. Appl. Phys. 71, 4933 ~1992!. 8 T. Lei, K. F. Ludwig, Jr., and T. D. Moustakas, J. Appl. Phys. 74, 4430 ~1993!. 9 H. Yang, O. Brandt, A. Trampert, and K. H. Ploog, Appl. Surf. Sci. 104/105, 461 ~1996!. 10 O. Brandt, H. Yang, A. Trampert, M. Wassermeier, and K. H. Ploog, Appl. Phys. Lett. 71, 473 ~1997!. 11 G. E. Crook, L. Da¨weritz, and K. Ploog, Phys. Rev. B 42, 5126 ~1990!. 12 S. N. Basu, T. Li, and T. D. Moustakas, J. Mater. Res. 9, 2370 ~1994!. 13 A. Ohtani, K. S. Stevens, and R. Beresford, Appl. Phys. Lett. 65, 61 ~1994!. 14 W. I. Wang, Appl. Phys. Lett. 44, 1149 ~1984!. 15 O. Brandt, H. Yang, B. Jenichen, Y. Suzuki, L. Da¨weritz, and K. H. Ploog, Phys. Rev. B 52, R2253 ~1995!. 16 O. Brandt, H. Yang, J. R. Mu¨llha¨user, A. Trampert, and K. H. Ploog, Mater. Sci. Eng. B 43, 214 ~1997!. 17 A. Trampert, O. Brandt, H. Yang, and K. H. Ploog, Appl. Phys. Lett. 70, 583 ~1997!. 18 J. Menniger, U. Jahn, O. Brandt, H. Yang, H. T. Grahn, and K. H. Ploog, Phys. Rev. B 53, 1881 ~1996!.

Downloaded 25 Jan 2013 to 62.141.165.1. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions