Growth evolution, adatom condensation, and island sizes in InGaAs/GaAs (001) R. Leon *, J. Wellman *, X. Z. Liao **, and J. Zou ** * Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, ** Australian Key Center for Microscopy and Microanalysis, Sydney, Australia MOCVD horizontal reactor cell operating at 1/10 atmosphere (CH 3 ) 3 Ga x10 -6, (CH 3 ) 3 In x10 -6, AsH x H 2 flow rate: 5.0 slm for graded growth [17.5 slm for uniform growth] Growth rate: 0.5 to 0.75 ml/sec Growth temperature: 550 o C Nominal composition: In 0.6 Ga 0.4 As Growth Rate Distance along GaAs Substrate (mm) Hydrogen flow (liters/minute) Graded Growth Reducing hydrogen flow results in a spatial gradient in growth rate. In our study the growth rate ranged from ml/sec. Uncapped Capped GaAs InGaAs GaAs InGaAs GaAs Schematic of specimens used in this study. Some were capped with 30 nm (un-graded) GaAs. QD Concentration 3. Variations in step density 4. Surface density variations. R. Leon and S. Fafard, PRB 58, R1726 (1998). As island concentration increases to saturation, average island diameter decreases. Compared capped and uncapped InGaAs/GaAs specimens. Philips 430 (300keV) and CM12 (120keV), using on-zone bright field imaging. Each frame below is 280 nm  280 nm. 5. Island size determination Capped Uncapped Both AFM (see frame 4) and TEM analysis [4-6] show that uncapped (low surface density) InGaAs islands are larger than equivalent islands after capping layer is grown. TEM images of uncapped islands show larger average island diameter at low concentration. TEM images of capped islands show no significant change in island diameter with concentration. 6. Island Sizes: capped vs. surface Size differences between surface dots and capped dots determined by AFM and plan view TEM imaging respectively. These measurements indicate that surface dots are larger, and that this size discrepancy increases with decreasing QD surface density. The layer-by-layer growth mode is observed for lattice- matched materials of identical crystal structure. Examples: Au on Ag and AlGaAs on GaAs. Van de Merwe Direct island growth is seen in materials with large lattice mismatch, high interfacial energy or different crystal structures. Examples: GaN on saphire and InAs on GaP. Volmer-Weber This type of growth occurs for crystals of dissimilar lattice parameters and low interfacial energy, like Ge on Si and InAs on GaAs. After an initial layer-by- layer growth, islands form spontaneously, leaving a thin “wetting layer” underneath. Stranski-Krastanow 1. Quantum Dot Formation In this work, we compare sizes of InGaAs islands (Quantum dots) before and after these are capped with the GaAs barrier. Large variations in InGaAs quantum dot concentrations were obtained with two different methods: (i) by simultaneous growths on vicinal GaAs [001] substrates with a range in surface step densities obtained by slight variations in substrate miscut angles [1]; and (ii) by using a graded growth rate which results in a positionally varying quantum dot density over a few millimeters [2]. Growth Modes: By choosing the appropriate materials, we can utilize the following growth modes to make Quantum Dots with defect free interfaces: + Variation in quantum dot concentration as a function of substrate miscut angle in simultaneous growths at 550  C. Measurements between capped and uncapped islands show good correspondence. Steps are energetically favorable sites for island nucleation; therefore, large variations in QD concentrations can be obtained by varying the availability of surface steps with growth conditions that minimize island coverage [3]. (a) From simultaneous growths on vicinal (001): (b) From graded growths: (d) Variations in InGaAs island concentrations as a function of ML coverage obtained for graded structures with a varying growth rate [2] for low (i) and high (ii) values of arsine partial pressure (5.7 x and 2.2 x respectively). Composite AFM deflection image showing island size evolution from low island concentration to island saturation. Above each 250- nm by 250-nm frame is a magnified image of a typical island. 7. Discussion Island diameters vs. average inter-dot separations for graded depositions. All filled (rose) symbols represent data for growths at low arsine partial pressure: squares are surface dots measured by AFM, diamonds are surface dots measured by TEM, and triangles are capped dots (TEM). Unfilled (white) squares correspond to surface islands formed at high arsine partial pressures. 8. Conclusions 9. References We suggest that the island size difference can be explained by group III thermal adatom condensation on existing islands during sample cooling. As shown in recent studies after STM analysis of rapidly quenched GaAs surfaces [7,8], Ga adatom concentrations on GaAs (001) surfaces are high (a significant fraction of a ML) at the growth temperatures normally used in Molecular Beam Epitaxy and MOCVD growth. This indicates that growth on GaAs (100) is closer to equilibrium than previously assumed. The islands act as a “sink” or “seed” for condensing adatoms, which migrate from the surface area surrounding each island. For widely spaced islands, migration lengths become a limitation, and this can explain the saturation in surface island dimensions seen in one of the growths shown in square 6. If we assume that migrating adatoms will condense on each island, the contributing surface will be distributed over a square area of side equal to the average inter-dot separation, minus the area occupied by the uncapped dot. Using reasonable values for “seed” island sizes (within the range of capped island sizes) and using thermal adatom concentrations obtained in [7], the growth of islands at intermediate concentrations can be accounted by condensing adatoms [9]. Differences in island coverage were obtained with simultaneous growths on different GaAs (100) vicinal surfaces and with structures deposited at growth rates varying with position. The size of surface and capped InGaAs QDs differ, and, This discrepancy increases with decreasing island coverage. There is good agreement between concentration values determined from capped and uncapped islands. We explain these size differences by condensing adatoms from different surface areas surrounding low and high-density islands. 1. R. Leon, S. Marcinkevicius, X. Z. Liao, J. Zou, D. J. H. Cockayne, and S. Fafard, Phys. Rev. B 60, R8517 (1999). 2. R. Leon and S. Fafard, Phys Rev. B 58, R1726 (1998). 3. R. Leon, C. Lobo, J. Zou, T. Romeo, and D. J. H. Cockayne, Phys. Rev. Lett. 81, 2486 (1998). 4. X. Z. Liao, J. Zou, X. F. Duan, D. J. H. Cockayne, R. Leon, and C. Lobo, Phys Rev B 58, R4235 (1998). 5. X. Z. Liao, J. Zou, D. J. H. Cockayne, R. Leon, and C. Lobo, Phys. Rev. Lett. 82, 5148 (1999). 6. J. Zou, X. Z. Liao, D. J. H. Cockayne and R. Leon, Phys. Rev B 59, 12,279 (1999). 7. M. D. Johnson, K. T. Leung, A. Birch, B. G. Orr, and J. Tersoff, Surf. Sci. 350, 254 (1996). 8. J. Tersoff, M. D. Johnson, and B. G. Orr, Phys. Rev. Lett. 78, 282 (1997). 9. R. Leon, J. Wellman, X. Z. Liao, J. Zou, and D. J. H. Cockayne, Appl. Phys. Lett 76, 1558 (2000). Once islands begin to form, island concentration increases exponentially to saturation in a fraction of a monolayer (i and ii correspond to different AsH 3 partial pressures). With a graded sample, we can study the continuous stages in island evolution on a single growth specimen. Differences in capped and uncapped island sizes are observed from different growth experiments. Surface islands, mainly for low concentrations, are significantly larger than their capped counterparts.