- Email: [email protected]
Self-assembled InAs quantum dots on anti-phase domains of GaAs on Ge substrates
Journal of Crystal Growth 323 (2011) 254–258
Contents lists available at ScienceDirect
Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro
Self-assembled InAs quantum dots on anti-phase domains of GaAs on Ge substrates W. Tantiweerasophon, S. Thainoi, P. Changmuang, S. Kanjanachuchai, S. Rattanathammaphan, S. Panyakeow n The Semiconductor Device Research Laboratory (SDRL), CoE Nanotechnology Center of Thailand, Department of Electrical Engineering, Faculty of Engineering, Chulalongkorn University, Phayathai Road, Pathumwan, Bangkok 10330, Thailand
a r t i c l e i n f o
a b s t r a c t
Available online 20 January 2011
The authors report the formation of self-assembled InAs quantum dots (QDs) grown on GaAs/Ge substrates having anti-phase domains (APDs) by molecular beam epitaxy. The AFM images of InAs QDs grown on different GaAs thicknesses are shown and compared. The samples with InAs coverage of 1.80 MLs with GaAs thickness of 300 and 700 nm show non-uniformed size distribution of the dots. Due to anisotropic property of quantum dots, ellipsoidal quantum dots appear. Unexpectedly, most of InAs quantum dots align perpendicularly to anti-phase boundary (APB) and to quantum dot alignment formed in adjacent domains. Photoluminescence spectrum excited by 20 mW 476-nm Ar + laser at 20 K does not show emission peak of InAs QDs. This is due to defects in the GaAs buffer layer. & 2011 Elsevier B.V. All rights reserved.
Keywords: A1. Atomic force microscopy A1. Low dimensional structures A1. Nanostructures A3. Molecular beam epitaxy B2. Semiconducting germanium B2. Semiconducting III–V materials
1. Introduction Lattice mismatch between GaAs and Ge is 0.1% leading to the possibility of forming defect-free heterostructure. Nevertheless, GaAs is a polar, III–V compound semiconductor whereas Ge is a non-polar, IV elemental semiconductor. The polar/non-polar heterointerface leads to anti-phase domains (APDs) in GaAs epitaxial layer grown on Ge substrates. The APDs can be eliminated using miscut Ge substrate [1]. In this work, we report the effect of the APDs and the crystalline quality of GaAs/Ge heteroepitaxial using migration-enhanced molecular beam epitaxy (MEE) [2] where group III and group V elements are separately deposited at different duration times. With suitable growth conditions, we could obtain larger GaAs crystal domains on Ge substrates. Combination of GaAs and Ge materials are applicable for many optoelectronic devices due to their bandgap properties, i.e. direct/indirect, wide/narrow bandgaps and their electronic properties, i.e. electron mobility, carrier life times. GaAs on Ge is a good candidate for tandem solar cells [3–5] in which GaAs part absorbs short wavelengths and Ge works for the remaining long wavelengths as the bottom cell. Tandem solar cells, therefore, have wide spectral response with good cell performance and cheap substrates. Many III–V compound semiconductors on Ge solar cells are widely used in space applications [6] and concentrated terrestrial systems [7].
n
Corresponding author. Tel.: +66 2 218 6521; fax: + 66 2 218 6523. E-mail addresses: [email protected], [email protected] (S. Panyakeow). 0022-0248/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2010.12.083
Self-assembled InAs quantum dots (QDs) grown on GaAs substrate are used as the active region of lasers [8–10] amplifiers, [11,12] infrared detectors, [13,14] and in other device applications. [15,16] Recently, Dhawan et al. [17] reported the growth of InAs quantum dots on germanium substrate using metal organic chemical vapor deposition and Knuuttila et al. [18] reported the growth of nonuniformly distributed InAs islands directly on Ge substrate without any buffer layer. In the growth of InAs QDs directly on Ge substrate, the formation of anti-phase domains is avoided since the dot formation occurs after a thin wetting layer is grown. Many defects will be in this thin wetting layer. This may adversely influence the growth of good quality dots later. Buffer layer assists in decreasing the surface defects of the substrate and help in confining InAs QDs. However, growth of uniformly distributed InAs QDs on standard (0 0 1) Ge substrates using GaAs buffer layer is rarely reported. InAs QDs are typically grown on miscut Ge substrates using GaAs buffer layer because APDs can be reduced or even eliminated using miscut Ge substrates. Motivated by the fact that on-axis Ge (0 0 1) substrates are cheaper and more widely used, we report the growth of InAs QDs on on-axis Ge (0 0 1) substrates using GaAs buffer layer and AlAs interfacial layer which proves to be effective as a blocking layer, preventing the cross diffusion of Ge, Ga and As atoms [19].
2. Experimental details Growth of GaAs, AlAs and InAs QDs are carried out using Riber’s 32P MBE machine. p-type on-axis Ge (0 0 1) substrates are used for all growth runs. Oxide removal and substrate degassing are accomplished by a 20 min 620 1C annealing in the growth
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
chamber having an in-situ reflection high energy electron diffraction (RHEED) monitoring. 30-nm AlAs interfacial layer is grown at 610 1C with a growth rate of 0.1 mm/h. The substrate temperature is ramped down to 350 1C and a two-step GaAs nucleation process is then employed. First, the initial 30-nm GaAs is deposited at 350 1C under the As4 flux of 5 10 5 T at a slow growth rate of 0.1 mm/h. Second, the subsequent growth is conducted at 350 1C with GaAs growth rate of 0.5 mm/h under As4 flux of 8 10 6 T. Film thicknesses for the two stages are 300 and 700 nm, respectively. After this MEE process is completed, the substrate temperature is ramped up to 500 1C and InAs QDs are grown with InAs coverage of 1.8 MLs. For photoluminescence (PL) measurements, the sample with 300-nm GaAs buffer layer is capped with GaAs, AlGaAs and GaAs layers at 510 1C with thicknesses of 50, 60 and 50 nm, respectively. Atomic force microscope (AFM) is used to study the surface morphology. Dot size, density and height were also determined as a function of GaAs buffer layer thickness. Optical characterizations of the samples were performed by PL measurements at 20 K with 476nm Ar + laser operated at 20 mW. The PL signal is detected by an InGaAs detector. The PL results and AFM images of the samples with 500 and 1000 nm GaAs buffer layer without InAs QDs are given for comparison.
3. Results and discussions When polar material (GaAs, AlAs and InAs) is epitaxially grown on a non-polar material (Ge), it always leads to structural defects called APDs, which create deep levels inside the forbidden band and acts as strong scattering centers [20,21]. These can also result in significant surface roughness and large area non-uniformity. The most common method for avoiding APDs from GaAs/Ge interfaces is using mis-oriented substrates [22–25]. The thickness of GaAs buffer layer is also crucial as APDs self-annihilate in a region of about 50 nm from the interface, leaving the GaAs final surface nearly free of APDs [24].
255
In order to reduce cost and the fact that oriented Ge (0 0 1) is practically used, we choose oriented Ge (0 0 1) substrate in our experiment. Figs. 1a and b show the AFM images of GaAs surface with thickness of 500 nm and 1 mm on p-Ge substrate. The AFM images show many APDs which result in anti-phase boundaries (APBs). The average domain size of the samples in Fig. 1b is approximately twice that shown in Fig. 1a and the depth of the domain boundaries is in the range 10–40 nm for both Figs. 1a and b. Domain size thus varies with GaAs thickness. The larger domain size indicates the lower APDs density, which in turn results in lower defects density, leading to better sample characteristic. However, the GaAs thickness should be optimized to provide suitable domain size for a good tandem solar cells. PL results of GaAs grown on AlAs/Ge substrate are shown in Figs. 2a and b with different measurement conditions. Fig. 2a shows the PL spectra of 500-nm GaAs grown on AlAs/Ge substrate with varying excitation laser power. From this figure, there is no spectral shift but only decreasing spectral peak intensity at 1.15 and 1.35 eV. All spectra of GaAs on AlAs/Ge show three different structures, one narrow band-to-band (B2B) at an energy of 1.5 eV, a broad inner-band-gap (IB) structure at an energy of 1.1 eV as also observed by Brammertz et al. [26] and unidentified structure at an energy of 1.3 eV. Small strain in the AlAs layer causes strain in GaAs buffer layer leading to the B2B structure, which leads to the separation into a light-hole and a heavy-hole peak. The IB structure observed from PL spectra consists of two Gaussian peaks with energies of 1.04 and 1.15 eV as shown in Fig. 3. This result is consistent with Brammertz’s result except for the peak at 1.35 eV. The peak at 1.04 eV is due to a GeGa–GeAs complex, whereas the peak at 1.17 eV is attributed to the GeGa–VGa complex [27]. One reasonable explanation from this result is that AlAs compound semiconductor is an ineffectively blocking layer if Ge substrate is oriented. APDs could be formed inside the AlAs structure, therefore Ge, Ga, As or even Al atoms can diffuse across AlAs/Ge interface and reach GaAs buffer layer leading to Ge-based complexes formed within GaAs buffer layer. More experimental
Fig. 1. AFM images of GaAs surface with thickness of (a) 500 nm and (b) 1 mm on p-Ge substrate.
256
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
Fig. 3. Two Gaussian peaks of IB structure with energy of 1.04 (a) and 1.15 eV (b) at 20 K and 40 mW. Unidentified structure and GaAs are at an energy of 1.3 (c) and 1.47 eV (d), respectively.
Fig. 2. PL spectra of 500 nm GaAs grown on AlAs/Ge substrate varying (a) laser power and (b) substrate temperature.
results are needed to explain this phenomena and to identify the spectral peak at 1.35 eV. From Fig. 2b, the PL peak is slightly red-shifted at high temperatures. This is reasonable because when the sample temperature is increased, existing splitting bands can merge, as predicted by Varshni equation [28]. QDs growth temperature and InAs coverage were fixed at 500 1C and 1.8 MLs at InAs growth rate of 0.01 ML/s. We use standard QDs growth conditions in which In and As shutters are simultaneously opened so there is a wetting layer (WL). Due to the high elastic potential energy, once certain critical thickness is achieved, additional atoms group together to form the quantum dot to reduce this elastic energy. The WL thickness is roughly 0.5 nm. Because of GaAs atoms placing on the wrong site resulting in APDs, the strain field significantly occurs around these APBs. This consequently impact on the WL of InAs. Before the InAs WL was nucleated, there has been the residual strain field inside GaAs, especially around the APBs. Thus, the WL critical thickness formed on APDs–GaAs is probably smaller than the WL critical thickness formed on GaAs without APDs. Fig. 4 shows the AFM images of InAs QDs grown on (a) 300-nm and (b) 700-nm GaAs/Ge. Histograms of QDs height distribution are also included in Fig. 4. In Stranski–Krastanov (SK) growth mode, few monolayers of twodimensional (2-D) growth appear prior to QD formation. The critical layer thickness at which this 2-D to 3-D transition occurs depends on the lattice mismatch between the layer being deposited and the GaAs buffer layer. For InAs layer on GaAs substrate, the critical layer thickness is around 1.6 MLs. As we increase
deposition thickness to 1.8 MLs which is above the critical thickness, QDs formation was observed, as shown in Fig. 4, with very low density of dots. Owing to anisotropic property of QDs, ellipsoid-shape quantum dots occur. Fig. 5 shows the cross section of QDs. The sample with GaAs thickness of 300 nm show size distribution of InAs dots having a length of major axis, minor axis and average height of 76.4 nm710%, 48.4 nm78% and 3.2 nm71% and the sample with GaAs thickness of 700 nm show size distribution of InAs dots having a length of major axis, minor axis and average height of 64.8 nm710%, 37.9 nm75% and 4.2 nm71%, respectively. Dot density, determined from the AFM images of Fig. 4, with GaAs thickness of 300 and 700 nm are 4.8 109 cm 2 and 5 109 cm 2, respectively. The elongated dots align in specific direction. Especially, in the case of large domain size, as shown in Fig. 4b, most of InAs quantum dots align in the direction normal to the boundary of APDs or APB and also to elongated quantum dot alignment formed in the adjacent domain. Each domain has only one specific crystallographic direction of the QDs elongation, as proved to be [1 1 0] direction by Suraprapapich et al. [29]. The orientation of GaAs zinc-blende structure and the single-step surface of (0 0 1) Ge [30] affect the strain field inside GaAs layer. Thus, QDs prefer to accumulate in the region possessing high strain field, which is the region near the APBs. This high strain field results from atoms placing on the wrong site. However, QDs are always aligned along [1 1 0] but, the [1 1 0] changes between the main phase and the other phase. The changing of domain orientation is the result of double domain reconstructed surfaces of (0 0 1) Ge [30]. The single steps on (0 0 1) Ge without miscut lead to double domain reconstructed surfaces of (0 0 1) Ge and also affect the GaAs domain orientation. The orientation of dimer bond of Ge atom on the single-step surface rotates 901 in each domain. As a result, the orientation of GaAs sublattice in each domain is also rotated by 901. We can use this elongated QDs alignment as indicator for crystallographic formation of each GaAs APDs. Fig. 6 shows the PL result of InAs QDs grown on 300 nm GaAs buffer layer with laser power of 20 mW at 20 K. We cannot observe any emission peaks of InAs QDs in this PL spectrum. This PL result is due to defects in GaAs buffer layer. Once electron in the QDs have been excited, the distribution of electron is no longer in equilibrium, and they eventually decay into lower states but, prior to reaching the appropriate lower states,
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
257
Fig. 4. AFM images of QDs grown on different GaAs thickness of (a) 300 nm and (b) 700 nm. Each inset picture shows histogram of QDs height distribution.
Fig. 5. 2D top-view AFM image and cross-sectional profiles of QDs grown on APDs–GaAs.
they are trapped by APB, which is formed closely to QDs alignment resulting in non-radiative recombination. Since APBs comprise planes of wrong nearest neighbor bonds where Ga atoms and As atoms reverse their positions crystallographically during the growth of polar material on the non-polar substrate, they are electrically active defects known to cause carrier scattering and non-radiative recombination [30,31]. Quantum dot quality can be improved by optimizing growth conditions, such as InAs coverage and QDs growth temperature, which consequently affect the improvement of PL results.
Fig. 6. PL result of InAs QDs grown on 300 nm GaAs buffer layer with laser power of 20 mW at 20 K.
Acknowledgements This research was supported by Nanotechnology Center of Thailand, Office of the High Education Commission, Thailand Research Fund, Asian Office of Aerospace Research and Development (AOARD) and Chulalongkorn University. The help and useful discussions about PL measurement with Dr. Noppadon Nuntawong are gratefully acknowledged.
4. Conclusions References Self-assembled InAs QDs are grown on Ge substrates using a GaAs buffer layer and AlAs interfacial layer. GaAs domain size varies with GaAs thickness. The effect of GaAs thickness on size, height and density of InAs QDs on Ge substrates are investigated. Anisotropic property affects the quantum dot shape. InAs QDs grown on APDs–GaAs buffer layer align in specific crystallographic direction for each domain. PL result does not show any emission peaks from InAs quantum dots.
[1] S.A. Ringel, R.M. Sieg, S.M. Ting, E.A. Fitzgerald, in: Proceedings of 26th PVSC, Anaheim, CA, September 30–October 3, 1997. [2] K.-Y. Chenh, IEEE Proc. 85 (11) (1997) 1694–1714. [3] K.A. Bertness, D.J. Friedman, Sarah R. Kurtz, A.E. Kibbler, C. Kramer, J.M. Olson, IEEE AES Syst. Mag. 9 (12) (1994) 12–17. [4] P. Loper, D. Hiller, M. Kunle, R. Gradmann, S. Janz, M. Hermle, M. Zacharias, S.W Glunz, Silicon quantum dot superstructures for all-silicon tandem solar cells: electrical andoptical characterization, /http://www.quantsol.org/pub/ pub09_31.pdfS.
258
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
[5] M.A. Green, G. Conibeer, I. Perez-Wurfl, S.J. Huang, D. Konig, D. Song, A. Gentle, X.J. Hao, S. Park, F. Gao, Y.H. So, Y. Huang, in: Proceedings of 23rd European Photovoltaic Solar Energy Conference and Exhibition, 1–5 September 2008, Valencia, Spain, pp. 1–4. [6] M.A. Stan, D.J. Aiken, P.R. Sharps, N.S. Fatemi, F.A. Spadafora, J. Hills, H. Yoo, B. Clevenger, in: Proceedings of 29th IEEE Photovoltaic Specialists Conference, 2002,, pp. 816–819. [7] S.R. Kurtz, D.J. Friedman, Presented at the National Center for Photovoltaics Program Review Meeting, Denver, Colorado, September 8–11, 1998. [8] K. Kamath, P. Bhattacharya, T. Sosnowski, T. Norris, J. Phillips, Electron. Lett. 32 (1996) 1374. [9] D.L. Huffaker, G. Park, Z. Zou, O.B. Shchekin, D.G. Deppe, Appl. Phys. Lett. 73 (1998) 2564. [10] A.R. Kovsh, et al., Electron. Lett. 35 (1999) 1161. [11] M. Sugawara, N. Hatori, T. Akiyama, Y. Nakata, H. Ishikawa, Jpn. J. Appl. Phys. 40 (2) (2001) 488. [12] P. Borri, et al., Appl. Phys. Lett. 79 (2001) 2633. [13] K.W. Berryman, S.A. Lyon, M. Segev, Appl. Phys. Lett. 70 (1997) 1861. [14] J. Philips, P. Bhattacharya, S.W. Kennerly, D.W. Beekman, M. Dutta, IEEE J. Quantum Electron. 35 (1999) 936. [15] S. Sangu, K. Kobayashi, A. Shojiguchi, M. Ohtsu, Phys. Rev. B 69 (2004) 115334. [16] A. Kiraz, M. Atature, A. Imamoglu, Phys. Rev. A 69 (2004) 032305.
[17] T. Dhawan, R. Tyagi, R.K. Bag, M. Singh, P. Mohan, T. Haldar, R. Murlidharan, R.P. Tandon, Nanoscale Res. Lett. 5 (2010) 31–37. [18] L. Knuuttila, K. Kainu, M. Sopanen, H. Lipsanen, J. Mater. Sci. Mater. Electron. 14 (5–7) (2003) 349. [19] C.K. Chia, J.R. Dong, D.Z. Chi, A. Sridhara, A.S.W. Wong, M. Suryana, G.K. Dalapati, S.J. Chua, S.J. Lee, Appl. Phys. Lett. 92 (2008) 141905. [20] H. Kroemer, J. Cryst. Growth 81 (1987) 193. [21] P.M. Petroff, J. Vac. Sci. Technol. B 4 (1986) 874. [22] Y. Li, L. Lazzarini, L.J. Giling, G. Salviati, J. Appl. Phys. 76 (1994) 5748. [23] L. Lazzarini, L. Nasi, G. Salviati, C.Z. Fregonara, Y. Li, L.J. Giling, C. Hardingham, D.B. Holt, Micron 31 (2000) 217. [24] W. Li, S. Laaksonen, J. Haapamaa, M. Pessa, J. Cryst. Growth 227–228 (2001) 104. [25] Y. Li, G. Salviati, M.M.G. Bongers, L. Lazzarini, L. Nasi, L.J. Giling, J. Cryst. Growth 163 (1996) 195. [26] G. Brammertz, Y. Mols, S. Degroote, V. Motsnyi, M. Leys, G. Borghs, M. Caymax, J. Appl. Phys. 99 (2006) 093514. [27] M.K. Hudait, P. Modak, S. Hardikar, S.B. Krupanidhi, J. Appl. Phys. 83 (1998) 4454. [28] K.P. O’Donnell, X. Chen, Appl. Phys. Lett. 58 (1991) 2924. [29] S. Suraprapapich, S. Kanjanachuchai, S. Thainoi, S. Panyakeow, Microelectron. Eng. 78–79 (2005) 349–352. [30] S.M. Ting, E.A. Fitzgerald, J. Appl. Phys. 87 (2000) 2618. [31] H. Noge, H. Kano, M. Hashimoto, I. Igarashi, J. Appl. Phys. 64 (1988) 2246.
Contents lists available at ScienceDirect
Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro
Self-assembled InAs quantum dots on anti-phase domains of GaAs on Ge substrates W. Tantiweerasophon, S. Thainoi, P. Changmuang, S. Kanjanachuchai, S. Rattanathammaphan, S. Panyakeow n The Semiconductor Device Research Laboratory (SDRL), CoE Nanotechnology Center of Thailand, Department of Electrical Engineering, Faculty of Engineering, Chulalongkorn University, Phayathai Road, Pathumwan, Bangkok 10330, Thailand
a r t i c l e i n f o
a b s t r a c t
Available online 20 January 2011
The authors report the formation of self-assembled InAs quantum dots (QDs) grown on GaAs/Ge substrates having anti-phase domains (APDs) by molecular beam epitaxy. The AFM images of InAs QDs grown on different GaAs thicknesses are shown and compared. The samples with InAs coverage of 1.80 MLs with GaAs thickness of 300 and 700 nm show non-uniformed size distribution of the dots. Due to anisotropic property of quantum dots, ellipsoidal quantum dots appear. Unexpectedly, most of InAs quantum dots align perpendicularly to anti-phase boundary (APB) and to quantum dot alignment formed in adjacent domains. Photoluminescence spectrum excited by 20 mW 476-nm Ar + laser at 20 K does not show emission peak of InAs QDs. This is due to defects in the GaAs buffer layer. & 2011 Elsevier B.V. All rights reserved.
Keywords: A1. Atomic force microscopy A1. Low dimensional structures A1. Nanostructures A3. Molecular beam epitaxy B2. Semiconducting germanium B2. Semiconducting III–V materials
1. Introduction Lattice mismatch between GaAs and Ge is 0.1% leading to the possibility of forming defect-free heterostructure. Nevertheless, GaAs is a polar, III–V compound semiconductor whereas Ge is a non-polar, IV elemental semiconductor. The polar/non-polar heterointerface leads to anti-phase domains (APDs) in GaAs epitaxial layer grown on Ge substrates. The APDs can be eliminated using miscut Ge substrate [1]. In this work, we report the effect of the APDs and the crystalline quality of GaAs/Ge heteroepitaxial using migration-enhanced molecular beam epitaxy (MEE) [2] where group III and group V elements are separately deposited at different duration times. With suitable growth conditions, we could obtain larger GaAs crystal domains on Ge substrates. Combination of GaAs and Ge materials are applicable for many optoelectronic devices due to their bandgap properties, i.e. direct/indirect, wide/narrow bandgaps and their electronic properties, i.e. electron mobility, carrier life times. GaAs on Ge is a good candidate for tandem solar cells [3–5] in which GaAs part absorbs short wavelengths and Ge works for the remaining long wavelengths as the bottom cell. Tandem solar cells, therefore, have wide spectral response with good cell performance and cheap substrates. Many III–V compound semiconductors on Ge solar cells are widely used in space applications [6] and concentrated terrestrial systems [7].
n
Corresponding author. Tel.: +66 2 218 6521; fax: + 66 2 218 6523. E-mail addresses: [email protected], [email protected] (S. Panyakeow). 0022-0248/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2010.12.083
Self-assembled InAs quantum dots (QDs) grown on GaAs substrate are used as the active region of lasers [8–10] amplifiers, [11,12] infrared detectors, [13,14] and in other device applications. [15,16] Recently, Dhawan et al. [17] reported the growth of InAs quantum dots on germanium substrate using metal organic chemical vapor deposition and Knuuttila et al. [18] reported the growth of nonuniformly distributed InAs islands directly on Ge substrate without any buffer layer. In the growth of InAs QDs directly on Ge substrate, the formation of anti-phase domains is avoided since the dot formation occurs after a thin wetting layer is grown. Many defects will be in this thin wetting layer. This may adversely influence the growth of good quality dots later. Buffer layer assists in decreasing the surface defects of the substrate and help in confining InAs QDs. However, growth of uniformly distributed InAs QDs on standard (0 0 1) Ge substrates using GaAs buffer layer is rarely reported. InAs QDs are typically grown on miscut Ge substrates using GaAs buffer layer because APDs can be reduced or even eliminated using miscut Ge substrates. Motivated by the fact that on-axis Ge (0 0 1) substrates are cheaper and more widely used, we report the growth of InAs QDs on on-axis Ge (0 0 1) substrates using GaAs buffer layer and AlAs interfacial layer which proves to be effective as a blocking layer, preventing the cross diffusion of Ge, Ga and As atoms [19].
2. Experimental details Growth of GaAs, AlAs and InAs QDs are carried out using Riber’s 32P MBE machine. p-type on-axis Ge (0 0 1) substrates are used for all growth runs. Oxide removal and substrate degassing are accomplished by a 20 min 620 1C annealing in the growth
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
chamber having an in-situ reflection high energy electron diffraction (RHEED) monitoring. 30-nm AlAs interfacial layer is grown at 610 1C with a growth rate of 0.1 mm/h. The substrate temperature is ramped down to 350 1C and a two-step GaAs nucleation process is then employed. First, the initial 30-nm GaAs is deposited at 350 1C under the As4 flux of 5 10 5 T at a slow growth rate of 0.1 mm/h. Second, the subsequent growth is conducted at 350 1C with GaAs growth rate of 0.5 mm/h under As4 flux of 8 10 6 T. Film thicknesses for the two stages are 300 and 700 nm, respectively. After this MEE process is completed, the substrate temperature is ramped up to 500 1C and InAs QDs are grown with InAs coverage of 1.8 MLs. For photoluminescence (PL) measurements, the sample with 300-nm GaAs buffer layer is capped with GaAs, AlGaAs and GaAs layers at 510 1C with thicknesses of 50, 60 and 50 nm, respectively. Atomic force microscope (AFM) is used to study the surface morphology. Dot size, density and height were also determined as a function of GaAs buffer layer thickness. Optical characterizations of the samples were performed by PL measurements at 20 K with 476nm Ar + laser operated at 20 mW. The PL signal is detected by an InGaAs detector. The PL results and AFM images of the samples with 500 and 1000 nm GaAs buffer layer without InAs QDs are given for comparison.
3. Results and discussions When polar material (GaAs, AlAs and InAs) is epitaxially grown on a non-polar material (Ge), it always leads to structural defects called APDs, which create deep levels inside the forbidden band and acts as strong scattering centers [20,21]. These can also result in significant surface roughness and large area non-uniformity. The most common method for avoiding APDs from GaAs/Ge interfaces is using mis-oriented substrates [22–25]. The thickness of GaAs buffer layer is also crucial as APDs self-annihilate in a region of about 50 nm from the interface, leaving the GaAs final surface nearly free of APDs [24].
255
In order to reduce cost and the fact that oriented Ge (0 0 1) is practically used, we choose oriented Ge (0 0 1) substrate in our experiment. Figs. 1a and b show the AFM images of GaAs surface with thickness of 500 nm and 1 mm on p-Ge substrate. The AFM images show many APDs which result in anti-phase boundaries (APBs). The average domain size of the samples in Fig. 1b is approximately twice that shown in Fig. 1a and the depth of the domain boundaries is in the range 10–40 nm for both Figs. 1a and b. Domain size thus varies with GaAs thickness. The larger domain size indicates the lower APDs density, which in turn results in lower defects density, leading to better sample characteristic. However, the GaAs thickness should be optimized to provide suitable domain size for a good tandem solar cells. PL results of GaAs grown on AlAs/Ge substrate are shown in Figs. 2a and b with different measurement conditions. Fig. 2a shows the PL spectra of 500-nm GaAs grown on AlAs/Ge substrate with varying excitation laser power. From this figure, there is no spectral shift but only decreasing spectral peak intensity at 1.15 and 1.35 eV. All spectra of GaAs on AlAs/Ge show three different structures, one narrow band-to-band (B2B) at an energy of 1.5 eV, a broad inner-band-gap (IB) structure at an energy of 1.1 eV as also observed by Brammertz et al. [26] and unidentified structure at an energy of 1.3 eV. Small strain in the AlAs layer causes strain in GaAs buffer layer leading to the B2B structure, which leads to the separation into a light-hole and a heavy-hole peak. The IB structure observed from PL spectra consists of two Gaussian peaks with energies of 1.04 and 1.15 eV as shown in Fig. 3. This result is consistent with Brammertz’s result except for the peak at 1.35 eV. The peak at 1.04 eV is due to a GeGa–GeAs complex, whereas the peak at 1.17 eV is attributed to the GeGa–VGa complex [27]. One reasonable explanation from this result is that AlAs compound semiconductor is an ineffectively blocking layer if Ge substrate is oriented. APDs could be formed inside the AlAs structure, therefore Ge, Ga, As or even Al atoms can diffuse across AlAs/Ge interface and reach GaAs buffer layer leading to Ge-based complexes formed within GaAs buffer layer. More experimental
Fig. 1. AFM images of GaAs surface with thickness of (a) 500 nm and (b) 1 mm on p-Ge substrate.
256
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
Fig. 3. Two Gaussian peaks of IB structure with energy of 1.04 (a) and 1.15 eV (b) at 20 K and 40 mW. Unidentified structure and GaAs are at an energy of 1.3 (c) and 1.47 eV (d), respectively.
Fig. 2. PL spectra of 500 nm GaAs grown on AlAs/Ge substrate varying (a) laser power and (b) substrate temperature.
results are needed to explain this phenomena and to identify the spectral peak at 1.35 eV. From Fig. 2b, the PL peak is slightly red-shifted at high temperatures. This is reasonable because when the sample temperature is increased, existing splitting bands can merge, as predicted by Varshni equation [28]. QDs growth temperature and InAs coverage were fixed at 500 1C and 1.8 MLs at InAs growth rate of 0.01 ML/s. We use standard QDs growth conditions in which In and As shutters are simultaneously opened so there is a wetting layer (WL). Due to the high elastic potential energy, once certain critical thickness is achieved, additional atoms group together to form the quantum dot to reduce this elastic energy. The WL thickness is roughly 0.5 nm. Because of GaAs atoms placing on the wrong site resulting in APDs, the strain field significantly occurs around these APBs. This consequently impact on the WL of InAs. Before the InAs WL was nucleated, there has been the residual strain field inside GaAs, especially around the APBs. Thus, the WL critical thickness formed on APDs–GaAs is probably smaller than the WL critical thickness formed on GaAs without APDs. Fig. 4 shows the AFM images of InAs QDs grown on (a) 300-nm and (b) 700-nm GaAs/Ge. Histograms of QDs height distribution are also included in Fig. 4. In Stranski–Krastanov (SK) growth mode, few monolayers of twodimensional (2-D) growth appear prior to QD formation. The critical layer thickness at which this 2-D to 3-D transition occurs depends on the lattice mismatch between the layer being deposited and the GaAs buffer layer. For InAs layer on GaAs substrate, the critical layer thickness is around 1.6 MLs. As we increase
deposition thickness to 1.8 MLs which is above the critical thickness, QDs formation was observed, as shown in Fig. 4, with very low density of dots. Owing to anisotropic property of QDs, ellipsoid-shape quantum dots occur. Fig. 5 shows the cross section of QDs. The sample with GaAs thickness of 300 nm show size distribution of InAs dots having a length of major axis, minor axis and average height of 76.4 nm710%, 48.4 nm78% and 3.2 nm71% and the sample with GaAs thickness of 700 nm show size distribution of InAs dots having a length of major axis, minor axis and average height of 64.8 nm710%, 37.9 nm75% and 4.2 nm71%, respectively. Dot density, determined from the AFM images of Fig. 4, with GaAs thickness of 300 and 700 nm are 4.8 109 cm 2 and 5 109 cm 2, respectively. The elongated dots align in specific direction. Especially, in the case of large domain size, as shown in Fig. 4b, most of InAs quantum dots align in the direction normal to the boundary of APDs or APB and also to elongated quantum dot alignment formed in the adjacent domain. Each domain has only one specific crystallographic direction of the QDs elongation, as proved to be [1 1 0] direction by Suraprapapich et al. [29]. The orientation of GaAs zinc-blende structure and the single-step surface of (0 0 1) Ge [30] affect the strain field inside GaAs layer. Thus, QDs prefer to accumulate in the region possessing high strain field, which is the region near the APBs. This high strain field results from atoms placing on the wrong site. However, QDs are always aligned along [1 1 0] but, the [1 1 0] changes between the main phase and the other phase. The changing of domain orientation is the result of double domain reconstructed surfaces of (0 0 1) Ge [30]. The single steps on (0 0 1) Ge without miscut lead to double domain reconstructed surfaces of (0 0 1) Ge and also affect the GaAs domain orientation. The orientation of dimer bond of Ge atom on the single-step surface rotates 901 in each domain. As a result, the orientation of GaAs sublattice in each domain is also rotated by 901. We can use this elongated QDs alignment as indicator for crystallographic formation of each GaAs APDs. Fig. 6 shows the PL result of InAs QDs grown on 300 nm GaAs buffer layer with laser power of 20 mW at 20 K. We cannot observe any emission peaks of InAs QDs in this PL spectrum. This PL result is due to defects in GaAs buffer layer. Once electron in the QDs have been excited, the distribution of electron is no longer in equilibrium, and they eventually decay into lower states but, prior to reaching the appropriate lower states,
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
257
Fig. 4. AFM images of QDs grown on different GaAs thickness of (a) 300 nm and (b) 700 nm. Each inset picture shows histogram of QDs height distribution.
Fig. 5. 2D top-view AFM image and cross-sectional profiles of QDs grown on APDs–GaAs.
they are trapped by APB, which is formed closely to QDs alignment resulting in non-radiative recombination. Since APBs comprise planes of wrong nearest neighbor bonds where Ga atoms and As atoms reverse their positions crystallographically during the growth of polar material on the non-polar substrate, they are electrically active defects known to cause carrier scattering and non-radiative recombination [30,31]. Quantum dot quality can be improved by optimizing growth conditions, such as InAs coverage and QDs growth temperature, which consequently affect the improvement of PL results.
Fig. 6. PL result of InAs QDs grown on 300 nm GaAs buffer layer with laser power of 20 mW at 20 K.
Acknowledgements This research was supported by Nanotechnology Center of Thailand, Office of the High Education Commission, Thailand Research Fund, Asian Office of Aerospace Research and Development (AOARD) and Chulalongkorn University. The help and useful discussions about PL measurement with Dr. Noppadon Nuntawong are gratefully acknowledged.
4. Conclusions References Self-assembled InAs QDs are grown on Ge substrates using a GaAs buffer layer and AlAs interfacial layer. GaAs domain size varies with GaAs thickness. The effect of GaAs thickness on size, height and density of InAs QDs on Ge substrates are investigated. Anisotropic property affects the quantum dot shape. InAs QDs grown on APDs–GaAs buffer layer align in specific crystallographic direction for each domain. PL result does not show any emission peaks from InAs quantum dots.
[1] S.A. Ringel, R.M. Sieg, S.M. Ting, E.A. Fitzgerald, in: Proceedings of 26th PVSC, Anaheim, CA, September 30–October 3, 1997. [2] K.-Y. Chenh, IEEE Proc. 85 (11) (1997) 1694–1714. [3] K.A. Bertness, D.J. Friedman, Sarah R. Kurtz, A.E. Kibbler, C. Kramer, J.M. Olson, IEEE AES Syst. Mag. 9 (12) (1994) 12–17. [4] P. Loper, D. Hiller, M. Kunle, R. Gradmann, S. Janz, M. Hermle, M. Zacharias, S.W Glunz, Silicon quantum dot superstructures for all-silicon tandem solar cells: electrical andoptical characterization, /http://www.quantsol.org/pub/ pub09_31.pdfS.
258
W. Tantiweerasophon et al. / Journal of Crystal Growth 323 (2011) 254–258
[5] M.A. Green, G. Conibeer, I. Perez-Wurfl, S.J. Huang, D. Konig, D. Song, A. Gentle, X.J. Hao, S. Park, F. Gao, Y.H. So, Y. Huang, in: Proceedings of 23rd European Photovoltaic Solar Energy Conference and Exhibition, 1–5 September 2008, Valencia, Spain, pp. 1–4. [6] M.A. Stan, D.J. Aiken, P.R. Sharps, N.S. Fatemi, F.A. Spadafora, J. Hills, H. Yoo, B. Clevenger, in: Proceedings of 29th IEEE Photovoltaic Specialists Conference, 2002,, pp. 816–819. [7] S.R. Kurtz, D.J. Friedman, Presented at the National Center for Photovoltaics Program Review Meeting, Denver, Colorado, September 8–11, 1998. [8] K. Kamath, P. Bhattacharya, T. Sosnowski, T. Norris, J. Phillips, Electron. Lett. 32 (1996) 1374. [9] D.L. Huffaker, G. Park, Z. Zou, O.B. Shchekin, D.G. Deppe, Appl. Phys. Lett. 73 (1998) 2564. [10] A.R. Kovsh, et al., Electron. Lett. 35 (1999) 1161. [11] M. Sugawara, N. Hatori, T. Akiyama, Y. Nakata, H. Ishikawa, Jpn. J. Appl. Phys. 40 (2) (2001) 488. [12] P. Borri, et al., Appl. Phys. Lett. 79 (2001) 2633. [13] K.W. Berryman, S.A. Lyon, M. Segev, Appl. Phys. Lett. 70 (1997) 1861. [14] J. Philips, P. Bhattacharya, S.W. Kennerly, D.W. Beekman, M. Dutta, IEEE J. Quantum Electron. 35 (1999) 936. [15] S. Sangu, K. Kobayashi, A. Shojiguchi, M. Ohtsu, Phys. Rev. B 69 (2004) 115334. [16] A. Kiraz, M. Atature, A. Imamoglu, Phys. Rev. A 69 (2004) 032305.
[17] T. Dhawan, R. Tyagi, R.K. Bag, M. Singh, P. Mohan, T. Haldar, R. Murlidharan, R.P. Tandon, Nanoscale Res. Lett. 5 (2010) 31–37. [18] L. Knuuttila, K. Kainu, M. Sopanen, H. Lipsanen, J. Mater. Sci. Mater. Electron. 14 (5–7) (2003) 349. [19] C.K. Chia, J.R. Dong, D.Z. Chi, A. Sridhara, A.S.W. Wong, M. Suryana, G.K. Dalapati, S.J. Chua, S.J. Lee, Appl. Phys. Lett. 92 (2008) 141905. [20] H. Kroemer, J. Cryst. Growth 81 (1987) 193. [21] P.M. Petroff, J. Vac. Sci. Technol. B 4 (1986) 874. [22] Y. Li, L. Lazzarini, L.J. Giling, G. Salviati, J. Appl. Phys. 76 (1994) 5748. [23] L. Lazzarini, L. Nasi, G. Salviati, C.Z. Fregonara, Y. Li, L.J. Giling, C. Hardingham, D.B. Holt, Micron 31 (2000) 217. [24] W. Li, S. Laaksonen, J. Haapamaa, M. Pessa, J. Cryst. Growth 227–228 (2001) 104. [25] Y. Li, G. Salviati, M.M.G. Bongers, L. Lazzarini, L. Nasi, L.J. Giling, J. Cryst. Growth 163 (1996) 195. [26] G. Brammertz, Y. Mols, S. Degroote, V. Motsnyi, M. Leys, G. Borghs, M. Caymax, J. Appl. Phys. 99 (2006) 093514. [27] M.K. Hudait, P. Modak, S. Hardikar, S.B. Krupanidhi, J. Appl. Phys. 83 (1998) 4454. [28] K.P. O’Donnell, X. Chen, Appl. Phys. Lett. 58 (1991) 2924. [29] S. Suraprapapich, S. Kanjanachuchai, S. Thainoi, S. Panyakeow, Microelectron. Eng. 78–79 (2005) 349–352. [30] S.M. Ting, E.A. Fitzgerald, J. Appl. Phys. 87 (2000) 2618. [31] H. Noge, H. Kano, M. Hashimoto, I. Igarashi, J. Appl. Phys. 64 (1988) 2246.