Fig. 1 - Semantic Scholar
Microelectronics Journal 35 (2004) 7–11 www.elsevier.com/locate/mejo
In segregation effects during quantum dot and quantum ring formation on GaAs(001) Jorge M. Garcı´a*, Daniel Granados, Juan Pedro Silveira, Fernando Briones Instituto de Microelectro´nica de Madrid, CNM-CSIC, C./Isaac Newton, 8, Tres Cantos, 28760 Madrid, Spain
Abstract In segregation during InAs growth on GaAs(001) is studied using a real time, in situ technique capable of measuring sample accumulated stress. A 50% surface In segregation of liquid-like stress free matter is deduced. A picture of growth below critical thickness for quantum dot formation is discussed on the basis of the equilibrium between pseudomorphic InAs and liquid In dominated by the stress energy. Quantum rings are produced when large (. 10 nm height) quantum dots are covered with 2 nm of GaAs cap. A formation mechanism of the rings is presented. The possibility of tailoring photoluminescence emission through control over size and shape is demonstrated. q 2004 Elsevier Ltd. All rights reserved. PACS: 81.15.Hi; 68.65. þ g; 64.75. þ 75 Keywords: Quantum dot; Quantum ring; Molecular beam epitaxy
InAs quantum dots (QD) are one of the most studied selfassembled systems. The physical origin of dot formation is the relaxation of the strain energy accumulated during the epitaxial growth of InAs onto another buffer material with a smaller lattice parameter. In spite of the importance of stress, little studies have been attempted to perform in situ, in real time characterization addressing the dot formation measuring the accumulated stress. Moreover, indium segregation effects in QD can lead also to changes of shape and composition [1], influencing the energy levels of the confined states [2]. A precise understanding of these processes can be used to control the size and the shape of QD [1]. In particular, a powerful technique to obtain self assembled quantum rings (QR) is based on the morphological changes of QD grown by molecular beam epitaxy (MBE) due to a thin GaAs capping [3 –5]. The purpose of this work is to present for the first time results on the evidence of stress induced liquid In together with a discussion for QR formation. We will show first a study of strain relaxation and In segregation effects during InAs growth on GaAs(001) both below and above critical thickness for QD formation. We use a technique capable to quantify accumulated stress measuring the deflection of a laser beam by a cantilever-shaped sample. Stress relaxation due to QD formation is observed and will be discussed. We * Corresponding author. Fax: þ 34-1-8060701. E-mail address: [email protected] (J.M. Garcı´a). 0026-2692/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/S0026-2692(03)00212-X
propose a complex growth mode of InAs on GaAs in which stress energy induces In surface segregation. We also present results of QR formation and will show that the final morphology and optical properties depend strongly on the starting QD shape, and, on detailed growth conditions such as the substrate temperature for the capping process ðTCAP Þ and the As molecular species flux. In situ, real time measurements of accumulated stress during heteroepitaxial MBE growth are performed by direct determination of strain induced substrate curvature using a laser deflection technique [6,7] along ½110 and ½110 directions. A detailed description of the technique can be found somewhere [8]. The substrate temperature ðTÞ is carefully calibrated by the observation on the reflection high energy electron diffraction (RHEED) diffraction pattern the oxide desorption (, 630 8C) and the surface reconstruction phase transition from a cð4 £ 4Þ to a clear 2 £ 4 [9]. During the experiment, the sample is continuously exposed to an As4 beam. InAs growth rate used was calibrated using RHEED oscillations. The QDs are grown by the deposition of 1.7 InAs monolayers (ML) which results in a low density (108 – 109cm22). For the ring formation, a GaAs thin capping layer is grown (1ML/s) at TCAP and annealed during 1 min. The samples for atomic force microscopy (AFM) characterization are cooled down immediately and removed from the chamber. The samples for photoluminescence spectroscopy (PL) measurements are obtained by burying these transformed nanostructures by a thick GaAs
8
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11
layer where the first 20 nm of GaAs are grown at TCAP and then T is increased to 595 8C. The final sample morphology has been characterized by ex situ contact mode AFM. Their emission characteristics are measured by PL using the 514.5 nm Ar laser line and a 0.22 m monochromator with a liquid nitrogen cooled Ge diode. For the study of In segregation, two kinds of structures have been grown on the cantilevers under real time observation of stress evolution: single monolayer (SM) and QD samples. For SM samples, a dose of In atoms sufficient to grow one nominal monolayer of InAs was supplied at different Ts from 200 to 500 8C. Detailed results on SM samples are reported elsewhere [10]. For QD samples, InAs is supplied up to doses slightly higher than the required critical thickness for QD formation (HQD). RHEED technique was used to monitor the 2D to 3D transition. Subsequent GaAs capping layer growth was initiated at the same T for QD formation during deposition of the first 10 nm and then Ts was ramped to 585 8C in order to obtain high quality GaAs. For the study of the ring formation, a second set of samples have been grown in which a growth pause of 60 s is introduced after the dots are capped with 2 nm of GaAs at 1ML/s at various T and As pressure. In all studied SM samples, during deposition of 1 equiv. ML of In, after a brief initial transient, related to surface reconstruction changes [6], an approximately linear increase of accumulated stress is observed. Except for this initial transient behavior, observed stress evolution is isotropic, cut cantilevers, as expected for identical for ½110 and ½110 an isotropic mismatched layer. Fig. 1 shows accumulated stress evolution at Ts ¼ 470 8C: A flat plateau is observed during a subsequent 30 s growth interruption. We do not expect any In desorption from the surface at this growth
temperature. Furthermore, during capping of InAs with GaAs, we observe again a progressive increase of accumulated stress until a final steady stress state (H1ML) is reached for a GaAs layer thickness always below 10 nm for all the studied growth conditions. This results can be explained assuming that during indium deposition, only a fraction of the delivered In is incorporated as InAs. The rest accumulates on the surface not contributing to the increase in stress. We want to point out that although both In and As have been supplied together, the actual accumulated stress due to growth of pseudomorphic InAs is approximately half of the expected (dashed line, Fig. 1). During subsequent GaAs capping deposition, this non-reacted In is progressively incorporated until total In exhaustion. The fraction of floating to In incorporated InAs can be evaluated from the ratio between partial accumulated stress achieved during InAs initial deposition and total accumulated stress induced after capping with a thick GaAs layer. For a wide range of growth temperatures [10] (from 200 to 500 8C) the measured total accumulated stress for one nominal monolayer of InAs reaches a value of H1ML ¼ 2.2 ^ 0.2 N/m. We also find that for temperatures commonly used to grow QD ðTs . 450 8CÞ approximately 50% of the supplied In does not incorporate actually into an InAs wetting layer (WL). It only incorporates progressively later during GaAs capping growth. This process shows that In segregation is not controlled only by a kinetic surface exchange process between III column atoms but mainly by a thermody-namic equilibrium controlled by the large surface stress energy associated to pseudomorphic InAs incorporation. The segregation profile during GaAs capping is determined by the progressive incorporation and depletion of this floating In under a combination of both thermodynamic equilibrium and growth kinetics mechanisms.
Fig. 1. Accumulated stress evolution during 1 In ML þ As4 deposition, growth interruption and subsequent GaAs capping. Total accumulated stress is ,2.25 N/m. Straight line corresponds to the expected accumulated stress increase rate if all supplied In were incorporating as InAs.
Fig. 2. Accumulated stress evolution during 2.3 In ML þ As4 deposition, growth interruption and subsequent GaAs capping. A clear reduction of stress increase rate is observed at point A due to QD formation, in coincidence with the appearance of a 3D RHEED pattern.
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11
9
Fig. 3. 1 nm £ 1 nm AFM images of: (a) QD grown at 540 8C, (b) QD after 2 nm capping at 500 8C with As2. The profile of the QD (c) and the rings (d) shows a drastic reduction in shape and height.
The large fraction (, 50%) of segregated In strongly modifies the standard picture for QD self-assembling process. Fig. 2 shows accumulated stress evolution during continuous supply of 2.3ML of InAs at 500 8C. A 50% reduction of stress increase rate is observed at point A corresponding to HQD ¼ 1.4ML. A 3D RHEED pattern appears at that moment. We have observed systematically a critical thickness for QD formation of 1.4 –1.7ML, in good agreement with other works [11]. When the QD are formed, the surface must be in a quasi-equilibrium among three phases: 2D InAs islands, 3D islands and liquid In. Due to the large lattice mismatch of InAs/GaAs (7.2%), strain energy competes efficiently with chemical bonding energy, resulting in a large surface In supersaturation. This competition determines the equilibrium ratio between bonded (InAs) and unbonded In. The appearance of QDs (with smaller associated stress accumulation due to surface relaxation) displaces previous equilibrium and reduces the observed accumulated stress slope as a function of delivered In (Fig. 2, point A). However, quantitative analysis of stress relaxation due to QD formation is complicated by mass transfer process among rapidly migrating surface In atoms, InAs 2D islands and QDs. The large increase of stress observed during GaAs capping (Fig. 2) is due mainly to the incorporation of In, but its final value cannot be directly and quantitatively related to the amount of this In. Now we should take into account the inhomogeneous nature of the stress field associated with the buried dots. On the other hand, Ga atoms supply during capping layer growth strongly modifies equilibrium between InAs and liquid In (or InAs) [14] as will be clearly shown later. During capping of InAs, In atoms will incorporate into the matrix in the form of an Inx Ga12x As alloy with a gradually decreasing x composition directly measurable with the cantilever technique. It is well known that the size and shape of the dots is modified when the InAs QDs are capped. Furthermore, In – Ga intermixing has been identified to be responsible for QD size change during capping [1]. This was already
observed in early dot experiments in which was clear that measured AFM height of uncapped QD was larger that the one obtained in capped QD samples by cross section transmission electron microscopy analysis. A drastic example of the effects of the capping process on the final shape of the QD is the QR formation. Fig. 3a shows a AFM image of InAs QD on GaAs. The vertical size distribution height of the dots is centered at 11 nm with a FWMH of 9%. Fig. 3c shows a profile plot of a typical QD. When a growth pause is introduced after a 2 nm GaAs cap is deposited on these dots at 500 8C under 2 £ 1026 mbar As2BEP each of them is transformed into one QR (Fig. 3b) with in-plane dimensions 100 nm £ 90 nm and , 1.5 nm high (Fig. 3d). It is worth mentioning that we never obtained QRs for samples with QD with a height less than 7 nm. We will focus on the mechanism for ring formation. Some authors have suggested [5,12,13] that a change in surface free energy balance due to the thin capping layer creates an outward pointing force. This force brings about a material redistribution resulting in a ring shaped structure. We propose that this effect is enhanced for certain growth conditions by a strain driven eruption of InAs from the topcenter of the dot. As it has been discussed before, a biaxial epitaxial stress of InAs on GaAs(001) leads to a mixture of stress-free matter including phases of ‘floating’ liquid In and InAs [14]. We think that small dots never reshape into rings because the amount of relaxed stress due to the formation of 3D islands is much smaller, and therefore there is no stress induced melting when they are partially capped. The observation of QR formation under As2 is related to an enhancement of the reactivity along the k110l direction [9]. The use of As2 should has similar effects than employing a larger As4 flux [15]. The effects of the reductions in vertical size are clearly observed in the PL spectra. Fig. 4, squares, shows PL emission from QD at 15 K. The inter-band transitions between ground and first excited state of electrons and holes are observed at low power excitation
10
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11
a competition of a T-independent de-wetting droplet instability process of liquid InAs and a strong T-dependent In –Ga alloying process. These morphological changes induces a reduction on the confining potential leading to modified energy levels as can be confirmed by means of PL measurements. QD have a highly efficient PL emission both at room and low (LT, i.e. 15 K) temperatures. Inter-band transitions between ground and first excited state of electrons and holes are observed at low power excitation (, 20 W/cm2), marked in Fig. 4 with an arrow. This allows the possibility of tailoring PL emission through control over size and shape.
Acknowledgements
Fig. 4. Low temperature PL spectra at 20 W/cm2 for samples with the nanostructures shown in Fig. 1 buried with GaAs. A strong blue shift in wavelength is observed from dots to rings emission.
(, 20 W/cm2). The PL emission of QR (Fig. 4, stars) shows a large blue shift of 220 meV with respect to the dots, suggesting a reduction of the vertical size associated with the formation of a central depleted region. In summary the Stranski – Krastanow model (in which after reaching a critical thickness the 2D grown layer breaks into 3D dots) does not seems to apply for a system where strong segregation effects are present as is the case for InAs on GaAs. A more complex picture arises from our direct stress measurements in which strong stress-induced In surface segregation controls mass transport and growth phenomena responsible for QD self-assembly. In segregation effects during 1ML InAs growth on GaAs(001) and InAs self-assembled QD formation are studied using a real time, in situ technique capable of measuring accumulated stress during growth. Due to a large (, 50%) surface In segregation of liquid-like stress free matter, self-assembled dot formation takes place when less than one monolayer of InAs is grown pseudomorphically on GaAs. A picture of growth process is discussed on the basis of the equilibrium between InAs and liquid In dominated by the stress energy in which large amount of mobile material is involved. The detailed understanding of these fundamental processes allows us to propose a formation mechanism of self assembled QR that are produced when large (11 nm height) QDs are covered with 2 nm of GaAs cap under certain growing conditions. The effect of partially cover a QD with GaAs is to destabilize the equilibrium forces shaping the QD, producing also a partial melting of the central region of the QD. Nanostructures of various shapes are obtained on changing the substrate temperature ðTÞ of the cap. The different material distribution can be understood as
The authors acknowledge David Go´mez for cantilever preparation. We would like to thank to Axel Lorke and Richard Warburton for fruitful discussions. This work has been partially supported by projects: 07T/0062/2000 of Comunidad Auto´noma de Madrid, TIC99-1035-C02, Nanoself TIC2002-04096 and by Nanomat of the EC Growth Program, contract no. G5RD-CT-2001-00545.
References [1] J.M. Garcı´a, G. Medeiros-Ribeiro, K. Schmidt, T. Ngo, P.M. Petroff, Intermixing and shape changes during the formation of InAs selfassembled quantum dots, Appl. Phys. Lett. 71 (1997) 2014– 2016. R. Songmuang, S. Kiravittaya, O.G. Schmidt, Shape evolution of InAs quantum dots during overgrowth, J. Cryst. Growth 249 (2003) 416 –417. [2] J.M. Garcı´a, T. Mankand, T. Holtz, P.J. Wellmann, P.M. Petroff, Electronic states tuning of InAs self-assembled quantum dots, Appl. Phys. Lett. 72 (1998) 3172–3174. [3] A. Lorke, R.J. Luyken, A.O. Govorov, J. Kotthaus, J.M. Garcı´a, P.M. Petroff, Spectroscopy of nanoscopic semiconductor rings, Phys. Rev. Lett. 84 (2000) 2223–2226. [4] R.J. Warburton, C. Schaflein, D. Haft, F. Bickel, A. Lorke, K. Karrai, J.M. Garcı´a, W. Schoenfeld, P.M. Petroff, Optical emission from single, charge-tunable quantum rings, Nature 405 (2000) 926 –927. [5] T. Raz, D. Ritter, G. Bahir, Formation of InAs self-assembled quantum rings on InP, Appl. Phys. Lett. 82 (2003) 1706. [6] J.P. Silveira, F. Briones, In situ observation of reconstruction related surface stress during molecular beam epitaxy (MBE) growth of III–V compounds, J. Cryst. Growth 201 (1999) 113–115. [7] J.A. Floro, E. Chason, S.R. Lee, R.D. Twesten, R.Q. Hwang, L.B. Freund, Real-time stress evolution during Si12x Gex heteroepitaxy: dislocations, islanding, and segregation, J. Electron. Mater. 26 (1997) 969 –971. [8] J.M. Garcı´a, J. Silveira, F. Briones, Strain relaxation and segregation during self assembled InAs quantum dots formation on GaAs(001), Appl. Phys. Lett. 77 (2000) 409–411. [9] D. Granados, J.M. Garcia, In(Ga)As self assembled quantum ring formation by molecular beam epitaxy, Appl. Phys. Lett. 82 (2003) 2401–2403. [10] J.P. Silveira, J.M. Garcia, F. Briones, Surface stress effects during MBE growth of III – V semiconductor nanostructures, J. Cryst. Growth 227 –228 (2001) 995– 999.
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11 [11] D. Leonard, K. Pond, P.M. Petroff, Critical thickness for selfassembled InAs islands on GaAs, Phys. Rev. B50 (1994) 11687. [12] A. Lorke, R. Blossey, J.M. Garcı´a, M. Bichler, G. Abstreiter, Morphological transformation of InGaAs islands, fabricated by Stranski–Krastanov growth, Mater. Sci. Engng B B88 (2002) 225–227. [13] R. Blossey, A. Lorke, Wetting droplet instability and quantum ring formation, Phys. Rev. E 65 (2002) 21603– 21605.
11
[14] D. Bottomley, The physical origin of InAs quantum dots on GaAs(001), Appl. Phys. Lett. 72 (1998) 783. D. Bottomley, Thermo-piezochemistry of InAs on GaAs(001), Appl. Phys. Lett. 80 (2002) 4747–4750. [15] T. Ogura, D. Kishimoto, T. Nishinaga, Effect of As molecular species on inter-surface diffusion in GaAs MBE for ridge structure fabrication, J. Cryst. Growth 226 (2001) 179 –182.
In segregation effects during quantum dot and quantum ring formation on GaAs(001) Jorge M. Garcı´a*, Daniel Granados, Juan Pedro Silveira, Fernando Briones Instituto de Microelectro´nica de Madrid, CNM-CSIC, C./Isaac Newton, 8, Tres Cantos, 28760 Madrid, Spain
Abstract In segregation during InAs growth on GaAs(001) is studied using a real time, in situ technique capable of measuring sample accumulated stress. A 50% surface In segregation of liquid-like stress free matter is deduced. A picture of growth below critical thickness for quantum dot formation is discussed on the basis of the equilibrium between pseudomorphic InAs and liquid In dominated by the stress energy. Quantum rings are produced when large (. 10 nm height) quantum dots are covered with 2 nm of GaAs cap. A formation mechanism of the rings is presented. The possibility of tailoring photoluminescence emission through control over size and shape is demonstrated. q 2004 Elsevier Ltd. All rights reserved. PACS: 81.15.Hi; 68.65. þ g; 64.75. þ 75 Keywords: Quantum dot; Quantum ring; Molecular beam epitaxy
InAs quantum dots (QD) are one of the most studied selfassembled systems. The physical origin of dot formation is the relaxation of the strain energy accumulated during the epitaxial growth of InAs onto another buffer material with a smaller lattice parameter. In spite of the importance of stress, little studies have been attempted to perform in situ, in real time characterization addressing the dot formation measuring the accumulated stress. Moreover, indium segregation effects in QD can lead also to changes of shape and composition [1], influencing the energy levels of the confined states [2]. A precise understanding of these processes can be used to control the size and the shape of QD [1]. In particular, a powerful technique to obtain self assembled quantum rings (QR) is based on the morphological changes of QD grown by molecular beam epitaxy (MBE) due to a thin GaAs capping [3 –5]. The purpose of this work is to present for the first time results on the evidence of stress induced liquid In together with a discussion for QR formation. We will show first a study of strain relaxation and In segregation effects during InAs growth on GaAs(001) both below and above critical thickness for QD formation. We use a technique capable to quantify accumulated stress measuring the deflection of a laser beam by a cantilever-shaped sample. Stress relaxation due to QD formation is observed and will be discussed. We * Corresponding author. Fax: þ 34-1-8060701. E-mail address: [email protected] (J.M. Garcı´a). 0026-2692/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/S0026-2692(03)00212-X
propose a complex growth mode of InAs on GaAs in which stress energy induces In surface segregation. We also present results of QR formation and will show that the final morphology and optical properties depend strongly on the starting QD shape, and, on detailed growth conditions such as the substrate temperature for the capping process ðTCAP Þ and the As molecular species flux. In situ, real time measurements of accumulated stress during heteroepitaxial MBE growth are performed by direct determination of strain induced substrate curvature using a laser deflection technique [6,7] along ½110 and ½110 directions. A detailed description of the technique can be found somewhere [8]. The substrate temperature ðTÞ is carefully calibrated by the observation on the reflection high energy electron diffraction (RHEED) diffraction pattern the oxide desorption (, 630 8C) and the surface reconstruction phase transition from a cð4 £ 4Þ to a clear 2 £ 4 [9]. During the experiment, the sample is continuously exposed to an As4 beam. InAs growth rate used was calibrated using RHEED oscillations. The QDs are grown by the deposition of 1.7 InAs monolayers (ML) which results in a low density (108 – 109cm22). For the ring formation, a GaAs thin capping layer is grown (1ML/s) at TCAP and annealed during 1 min. The samples for atomic force microscopy (AFM) characterization are cooled down immediately and removed from the chamber. The samples for photoluminescence spectroscopy (PL) measurements are obtained by burying these transformed nanostructures by a thick GaAs
8
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11
layer where the first 20 nm of GaAs are grown at TCAP and then T is increased to 595 8C. The final sample morphology has been characterized by ex situ contact mode AFM. Their emission characteristics are measured by PL using the 514.5 nm Ar laser line and a 0.22 m monochromator with a liquid nitrogen cooled Ge diode. For the study of In segregation, two kinds of structures have been grown on the cantilevers under real time observation of stress evolution: single monolayer (SM) and QD samples. For SM samples, a dose of In atoms sufficient to grow one nominal monolayer of InAs was supplied at different Ts from 200 to 500 8C. Detailed results on SM samples are reported elsewhere [10]. For QD samples, InAs is supplied up to doses slightly higher than the required critical thickness for QD formation (HQD). RHEED technique was used to monitor the 2D to 3D transition. Subsequent GaAs capping layer growth was initiated at the same T for QD formation during deposition of the first 10 nm and then Ts was ramped to 585 8C in order to obtain high quality GaAs. For the study of the ring formation, a second set of samples have been grown in which a growth pause of 60 s is introduced after the dots are capped with 2 nm of GaAs at 1ML/s at various T and As pressure. In all studied SM samples, during deposition of 1 equiv. ML of In, after a brief initial transient, related to surface reconstruction changes [6], an approximately linear increase of accumulated stress is observed. Except for this initial transient behavior, observed stress evolution is isotropic, cut cantilevers, as expected for identical for ½110 and ½110 an isotropic mismatched layer. Fig. 1 shows accumulated stress evolution at Ts ¼ 470 8C: A flat plateau is observed during a subsequent 30 s growth interruption. We do not expect any In desorption from the surface at this growth
temperature. Furthermore, during capping of InAs with GaAs, we observe again a progressive increase of accumulated stress until a final steady stress state (H1ML) is reached for a GaAs layer thickness always below 10 nm for all the studied growth conditions. This results can be explained assuming that during indium deposition, only a fraction of the delivered In is incorporated as InAs. The rest accumulates on the surface not contributing to the increase in stress. We want to point out that although both In and As have been supplied together, the actual accumulated stress due to growth of pseudomorphic InAs is approximately half of the expected (dashed line, Fig. 1). During subsequent GaAs capping deposition, this non-reacted In is progressively incorporated until total In exhaustion. The fraction of floating to In incorporated InAs can be evaluated from the ratio between partial accumulated stress achieved during InAs initial deposition and total accumulated stress induced after capping with a thick GaAs layer. For a wide range of growth temperatures [10] (from 200 to 500 8C) the measured total accumulated stress for one nominal monolayer of InAs reaches a value of H1ML ¼ 2.2 ^ 0.2 N/m. We also find that for temperatures commonly used to grow QD ðTs . 450 8CÞ approximately 50% of the supplied In does not incorporate actually into an InAs wetting layer (WL). It only incorporates progressively later during GaAs capping growth. This process shows that In segregation is not controlled only by a kinetic surface exchange process between III column atoms but mainly by a thermody-namic equilibrium controlled by the large surface stress energy associated to pseudomorphic InAs incorporation. The segregation profile during GaAs capping is determined by the progressive incorporation and depletion of this floating In under a combination of both thermodynamic equilibrium and growth kinetics mechanisms.
Fig. 1. Accumulated stress evolution during 1 In ML þ As4 deposition, growth interruption and subsequent GaAs capping. Total accumulated stress is ,2.25 N/m. Straight line corresponds to the expected accumulated stress increase rate if all supplied In were incorporating as InAs.
Fig. 2. Accumulated stress evolution during 2.3 In ML þ As4 deposition, growth interruption and subsequent GaAs capping. A clear reduction of stress increase rate is observed at point A due to QD formation, in coincidence with the appearance of a 3D RHEED pattern.
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11
9
Fig. 3. 1 nm £ 1 nm AFM images of: (a) QD grown at 540 8C, (b) QD after 2 nm capping at 500 8C with As2. The profile of the QD (c) and the rings (d) shows a drastic reduction in shape and height.
The large fraction (, 50%) of segregated In strongly modifies the standard picture for QD self-assembling process. Fig. 2 shows accumulated stress evolution during continuous supply of 2.3ML of InAs at 500 8C. A 50% reduction of stress increase rate is observed at point A corresponding to HQD ¼ 1.4ML. A 3D RHEED pattern appears at that moment. We have observed systematically a critical thickness for QD formation of 1.4 –1.7ML, in good agreement with other works [11]. When the QD are formed, the surface must be in a quasi-equilibrium among three phases: 2D InAs islands, 3D islands and liquid In. Due to the large lattice mismatch of InAs/GaAs (7.2%), strain energy competes efficiently with chemical bonding energy, resulting in a large surface In supersaturation. This competition determines the equilibrium ratio between bonded (InAs) and unbonded In. The appearance of QDs (with smaller associated stress accumulation due to surface relaxation) displaces previous equilibrium and reduces the observed accumulated stress slope as a function of delivered In (Fig. 2, point A). However, quantitative analysis of stress relaxation due to QD formation is complicated by mass transfer process among rapidly migrating surface In atoms, InAs 2D islands and QDs. The large increase of stress observed during GaAs capping (Fig. 2) is due mainly to the incorporation of In, but its final value cannot be directly and quantitatively related to the amount of this In. Now we should take into account the inhomogeneous nature of the stress field associated with the buried dots. On the other hand, Ga atoms supply during capping layer growth strongly modifies equilibrium between InAs and liquid In (or InAs) [14] as will be clearly shown later. During capping of InAs, In atoms will incorporate into the matrix in the form of an Inx Ga12x As alloy with a gradually decreasing x composition directly measurable with the cantilever technique. It is well known that the size and shape of the dots is modified when the InAs QDs are capped. Furthermore, In – Ga intermixing has been identified to be responsible for QD size change during capping [1]. This was already
observed in early dot experiments in which was clear that measured AFM height of uncapped QD was larger that the one obtained in capped QD samples by cross section transmission electron microscopy analysis. A drastic example of the effects of the capping process on the final shape of the QD is the QR formation. Fig. 3a shows a AFM image of InAs QD on GaAs. The vertical size distribution height of the dots is centered at 11 nm with a FWMH of 9%. Fig. 3c shows a profile plot of a typical QD. When a growth pause is introduced after a 2 nm GaAs cap is deposited on these dots at 500 8C under 2 £ 1026 mbar As2BEP each of them is transformed into one QR (Fig. 3b) with in-plane dimensions 100 nm £ 90 nm and , 1.5 nm high (Fig. 3d). It is worth mentioning that we never obtained QRs for samples with QD with a height less than 7 nm. We will focus on the mechanism for ring formation. Some authors have suggested [5,12,13] that a change in surface free energy balance due to the thin capping layer creates an outward pointing force. This force brings about a material redistribution resulting in a ring shaped structure. We propose that this effect is enhanced for certain growth conditions by a strain driven eruption of InAs from the topcenter of the dot. As it has been discussed before, a biaxial epitaxial stress of InAs on GaAs(001) leads to a mixture of stress-free matter including phases of ‘floating’ liquid In and InAs [14]. We think that small dots never reshape into rings because the amount of relaxed stress due to the formation of 3D islands is much smaller, and therefore there is no stress induced melting when they are partially capped. The observation of QR formation under As2 is related to an enhancement of the reactivity along the k110l direction [9]. The use of As2 should has similar effects than employing a larger As4 flux [15]. The effects of the reductions in vertical size are clearly observed in the PL spectra. Fig. 4, squares, shows PL emission from QD at 15 K. The inter-band transitions between ground and first excited state of electrons and holes are observed at low power excitation
10
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11
a competition of a T-independent de-wetting droplet instability process of liquid InAs and a strong T-dependent In –Ga alloying process. These morphological changes induces a reduction on the confining potential leading to modified energy levels as can be confirmed by means of PL measurements. QD have a highly efficient PL emission both at room and low (LT, i.e. 15 K) temperatures. Inter-band transitions between ground and first excited state of electrons and holes are observed at low power excitation (, 20 W/cm2), marked in Fig. 4 with an arrow. This allows the possibility of tailoring PL emission through control over size and shape.
Acknowledgements
Fig. 4. Low temperature PL spectra at 20 W/cm2 for samples with the nanostructures shown in Fig. 1 buried with GaAs. A strong blue shift in wavelength is observed from dots to rings emission.
(, 20 W/cm2). The PL emission of QR (Fig. 4, stars) shows a large blue shift of 220 meV with respect to the dots, suggesting a reduction of the vertical size associated with the formation of a central depleted region. In summary the Stranski – Krastanow model (in which after reaching a critical thickness the 2D grown layer breaks into 3D dots) does not seems to apply for a system where strong segregation effects are present as is the case for InAs on GaAs. A more complex picture arises from our direct stress measurements in which strong stress-induced In surface segregation controls mass transport and growth phenomena responsible for QD self-assembly. In segregation effects during 1ML InAs growth on GaAs(001) and InAs self-assembled QD formation are studied using a real time, in situ technique capable of measuring accumulated stress during growth. Due to a large (, 50%) surface In segregation of liquid-like stress free matter, self-assembled dot formation takes place when less than one monolayer of InAs is grown pseudomorphically on GaAs. A picture of growth process is discussed on the basis of the equilibrium between InAs and liquid In dominated by the stress energy in which large amount of mobile material is involved. The detailed understanding of these fundamental processes allows us to propose a formation mechanism of self assembled QR that are produced when large (11 nm height) QDs are covered with 2 nm of GaAs cap under certain growing conditions. The effect of partially cover a QD with GaAs is to destabilize the equilibrium forces shaping the QD, producing also a partial melting of the central region of the QD. Nanostructures of various shapes are obtained on changing the substrate temperature ðTÞ of the cap. The different material distribution can be understood as
The authors acknowledge David Go´mez for cantilever preparation. We would like to thank to Axel Lorke and Richard Warburton for fruitful discussions. This work has been partially supported by projects: 07T/0062/2000 of Comunidad Auto´noma de Madrid, TIC99-1035-C02, Nanoself TIC2002-04096 and by Nanomat of the EC Growth Program, contract no. G5RD-CT-2001-00545.
References [1] J.M. Garcı´a, G. Medeiros-Ribeiro, K. Schmidt, T. Ngo, P.M. Petroff, Intermixing and shape changes during the formation of InAs selfassembled quantum dots, Appl. Phys. Lett. 71 (1997) 2014– 2016. R. Songmuang, S. Kiravittaya, O.G. Schmidt, Shape evolution of InAs quantum dots during overgrowth, J. Cryst. Growth 249 (2003) 416 –417. [2] J.M. Garcı´a, T. Mankand, T. Holtz, P.J. Wellmann, P.M. Petroff, Electronic states tuning of InAs self-assembled quantum dots, Appl. Phys. Lett. 72 (1998) 3172–3174. [3] A. Lorke, R.J. Luyken, A.O. Govorov, J. Kotthaus, J.M. Garcı´a, P.M. Petroff, Spectroscopy of nanoscopic semiconductor rings, Phys. Rev. Lett. 84 (2000) 2223–2226. [4] R.J. Warburton, C. Schaflein, D. Haft, F. Bickel, A. Lorke, K. Karrai, J.M. Garcı´a, W. Schoenfeld, P.M. Petroff, Optical emission from single, charge-tunable quantum rings, Nature 405 (2000) 926 –927. [5] T. Raz, D. Ritter, G. Bahir, Formation of InAs self-assembled quantum rings on InP, Appl. Phys. Lett. 82 (2003) 1706. [6] J.P. Silveira, F. Briones, In situ observation of reconstruction related surface stress during molecular beam epitaxy (MBE) growth of III–V compounds, J. Cryst. Growth 201 (1999) 113–115. [7] J.A. Floro, E. Chason, S.R. Lee, R.D. Twesten, R.Q. Hwang, L.B. Freund, Real-time stress evolution during Si12x Gex heteroepitaxy: dislocations, islanding, and segregation, J. Electron. Mater. 26 (1997) 969 –971. [8] J.M. Garcı´a, J. Silveira, F. Briones, Strain relaxation and segregation during self assembled InAs quantum dots formation on GaAs(001), Appl. Phys. Lett. 77 (2000) 409–411. [9] D. Granados, J.M. Garcia, In(Ga)As self assembled quantum ring formation by molecular beam epitaxy, Appl. Phys. Lett. 82 (2003) 2401–2403. [10] J.P. Silveira, J.M. Garcia, F. Briones, Surface stress effects during MBE growth of III – V semiconductor nanostructures, J. Cryst. Growth 227 –228 (2001) 995– 999.
J.M. Garcı´a et al. / Microelectronics Journal 35 (2004) 7–11 [11] D. Leonard, K. Pond, P.M. Petroff, Critical thickness for selfassembled InAs islands on GaAs, Phys. Rev. B50 (1994) 11687. [12] A. Lorke, R. Blossey, J.M. Garcı´a, M. Bichler, G. Abstreiter, Morphological transformation of InGaAs islands, fabricated by Stranski–Krastanov growth, Mater. Sci. Engng B B88 (2002) 225–227. [13] R. Blossey, A. Lorke, Wetting droplet instability and quantum ring formation, Phys. Rev. E 65 (2002) 21603– 21605.
11
[14] D. Bottomley, The physical origin of InAs quantum dots on GaAs(001), Appl. Phys. Lett. 72 (1998) 783. D. Bottomley, Thermo-piezochemistry of InAs on GaAs(001), Appl. Phys. Lett. 80 (2002) 4747–4750. [15] T. Ogura, D. Kishimoto, T. Nishinaga, Effect of As molecular species on inter-surface diffusion in GaAs MBE for ridge structure fabrication, J. Cryst. Growth 226 (2001) 179 –182.
