pubs.acs.org/NanoLett
Band Alignment Tailoring of InAs1-xSbx/GaAs Quantum Dots: Control of Type I to Type II Transition J. He,* C. J. Reyner, B. L. Liang, K. Nunna, and D. L. Huffaker* California NanoSystems Institute, University of California, Los Angeles, California 90095
N. Pavarelli, K. Gradkowski, T. J. Ochalski, and G. Huyet Tyndall National Institute, University College Cork and Cork Institute of Technology, Cork, Republic of Ireland
V. G. Dorogan, Yu. I. Mazur, and G. J. Salamo Department of Physics, University of Arkansas, Fayetteville, Arkansas 72701 ABSTRACT We report the growth of InAs1-xSbx self-assembled quantum dots (QDs) on GaAs (100) by molecular beam epitaxy. The optical properties of the QDs are investigated by photoluminescence (PL) and time-resolved photoluminescence (TRPL). A type I to type II band alignment transition is demonstrated by both power-dependent PL and TRPL in InAs1-xSbx QD samples with increased Sb beam flux. Results are compared to an eight-band strain-dependent k·p model incorporating detailed QD structure and alloy composition. The calculations show that the conduction band offset of InAs1-xSbx/GaAs can be continuously tuned from 0 to 500 meV and a flat conduction band alignment exists when 60% Sb is incorporated into the QDs. Our study offers the possibility of tailoring the band structure of GaAs based InAsSb QDs and opens up new means for device applications. KEYWORDS Quantum dot, band alignment, optical properties
S
est in developing a material system with a flat conduction or valence band, which enables an undisturbed efficient charge transfer.18,19 Potential applications of these structures include thermoelectric devices, high efficiency solar cells, and lasers. However, the only two means of adjusting the band alignment of In-based QDs, including the abovementioned InAs/GaAs or InSb/GaAs QDs, have been with selected capping materials or with their incorporation into another quantum structure. Neither option is able to produce a flat band.7,8,11,12 The incorporation of Sb into In-based QDs on GaAs affords the advantage of tailoring the interfacial energy levels, offering both a flat-band condition and a new degree of freedom for tailoring the band structure. In this study, we report the formation of InAsSb QDs on GaAs(100). The type I to type II band alignment transition of InAsSb QDs is observed using both power-dependent photoluminescence (PL) and time-resolved photoluminescence (TRPL) studies. Simulations are performed to confirm the experimental results. Eight-band strain-dependent k·p calculations show that the conduction band offset can be continuously tuned from 0 to 500 meV, and most importantly, the flat conduction band structure (0 meV conduction band offset) of InAsSb/GaAs heterostructure can be achieved when Sb content is around 60%. This continuous tuning of conduction band offset and flat-band structure will lead to
elf-assembled quantum dots (QDs) have been extensively studied, and a good understanding of their electronic and optical properties has been achieved both experimentally and theoretically.1-12 Typically this research has been performed on type I InAs/GaAs QDs where electrons and holes are confined inside QDs simultaneously, leading to promising applications for optoelectronic devices, such as single photon QD sources and ultralow-threshold QD lasers.13-15 Historically, InSb/GaAs QDs have not received the same attention due to growth difficulties associated with Sb compounds and their type II band alignment.16,17 Contrary to type I, type II alignment acts as a potential barrier for one species of carrier, providing therefore confinement only for the other one. This spatial separation of charges makes these materials less suitable for emitters but more suitable in photovoltaic or photoconduction applications. A type I to type II band alignment transition is expected by varying the InAs1-xSbx composition from pure InAs to InSb. The transition point between the two alignments for InAs1-xSbx/GaAs QD has never been ascertained. Moreover, researchers have expressed strong inter-
* To whom correspondence should be addressed,
[email protected] and
[email protected]. Received for review: 05/3/2010 Published on Web: 07/22/2010 © 2010 American Chemical Society
3052
DOI: 10.1021/nl102237n | Nano Lett. 2010, 10, 3052–3056
potential device application, especially in high-efficiency solar cells and thermoelectric devices. The samples are grown by solid source molecular beam epitaxy (MBE) on semi-insulating GaAs(100) substrates. After native oxide desorption at 580 °C, a 100 nm GaAs buffer layer is grown at a substrate temperature of 580 °C. The temperature is then lowered to 480 °C for the deposition of InAsSb. The QDs are grown at 0.08 ML/s with InAsSb coverage of 2.1 monolayers (MLs). Four InAsSb QD samples (A, B, C, D) with varying Sb beam flux are grown. An InAs/ GaAs QD sample is also grown as a reference sample. In order to conduct atomic force microscopy (AFM), the growth is stopped after 2.1 ML deposition of InAsSb, and the samples are quickly cooled. For PL and TRPL measurements, a 10 nm GaAs capping layer is grown at 480 °C to avoid indium evaporation. The temperature is then increased to 580 °C, and an additional 90 nm GaAs layer is grown to cap the QDs. The Sb2 beam flux is 2.4 × 10-5, 4.0 × 10-5, 8.0 × 10-5, and 1.24 × 10-4 Pa for samples A, B, C, and D respectively. The As2 beam flux is maintained between 6.7 × 10-5 and 1.6 × 10-4 Pa for all samples. The realization of abrupt semiconductor heterointerfaces is inherently limited by surface segregation effects. Extensive studies have been carried on InAs/GaAs and GaSb/GaAs systems. In the InAs/ GaAs material system, the segregation coefficients of In have been reported in the range of 0.80-0.87,20,21 while in GaSb/ GaAs, the segregation coefficient of Sb is in the range of 0.6-0.95,22,23 strongly depending on the growth condition, especially the substrate temperature. The segregation, together with the capping process,4,5,24 leads to nonuniform In and Sb distribution within QDs and changes the buried QDs composition as compared to that of free-standing QDs, which affects the optical properties of the QDs. It is worth noting that, due to the incorporation of Sb, InAsSb/GaAs QD growth temperature needs to be relatively low when compared to that of InAs/GaAs QDs, which might suppress the intermixing during the capping process. During the growth of InAs1-xSbx QDs, the reflection highenergy electron diffraction (RHEED) pattern viewed along the [110] azimuth consists of well-defined spots which evolve after 1.5-1.6 ML InAsSb deposition, indicating the formation of InAsSb QDs. With higher Sb beam fluxes, the critical thickness is reduced from 1.7 ML (reference sample) to 1.5 ML (sample D) due to the increment of compressive strain. The AFM images of samples A and C are depicted in Figure 1. They show that the QD density is about 4 × 1010 cm-2 for sample A in Figure 1a and decreases to 1 × 1010 cm-2 with the increase of Sb beam flux to 8 × 10-5 Pa (sample C) shown in Figure 1b, whereas the diameter of the QDs is around 20 nm for both samples. We notice that a shape anisotropy of InAsSb QDs is clearly visible from both AFM images, and the lateral aspect ratio decreases with the Sb content increase from panel a to panel b of Figure 1. Marquez et al. studied atomically resolved structure of uncovered self-assembled InAs QDs based on atomically © 2010 American Chemical Society
FIGURE 1. AFM images of (a) sample A and (b) sample C.
FIGURE 2. Power-dependent PL spectra of (a) reference sample, (b) sample A, (c) sample C, and (d) sample D. The same color refers to the same excitation power for all the four graphs.
resolved scanning tunneling microscopy images and observed that InAs/GaAs QDs are bounded by at least 10 facets.4 Four of them determine the shape and are of {137} facets, which account for the observed shape anisotropy of InAs/GaAs QDs. We believe that the lateral aspect ratio in Figure 1 is related to the well-known {137} facets of uncapped InAs QDs getting decreased by the Sb content, where new and probably more symmetric surface reconstructions can evolve. Furthermore, the AFM usually measures higher lateral sizes; thus the diameter of the QDs is actually smaller than 20 nm. It is worth noting that, with increasing Sb content, some InAsSb islands tend to nucleate to consume the residual strain energy as shown for sample C in Figure 1b. Figure 2 shows the PL spectra of samples A, C, and D as well as the reference recorded at 77 K as a function of excitation power. For the InAs/GaAs reference sample, a strong emission peak is observed at 1130 nm as shown in Figure 2a. This signal originates from the radiative recombination in the ground state E0 of InAs QDs. As the excitation power is increased, excited states are filled from lower to higher energies indicated as E1, E2, and E3, while the ground state stays in the same position. This behavior is typical of type I QDs. It is well-known that a small amount of Sb is found to act as a surfactant, improving the wetting layer quality and increasing the critical thickness for QD formation by altering both the surface diffusion kinetics and the surface energy. Thus, larger QDs are obtained, which results in a red 3053
DOI: 10.1021/nl102237n | Nano Lett. 2010, 10, 3052-–3056
shift of the QD luminescence due to the change of quantum size effect. Heitz et al. reported that the Sb surfactant effect, together with extended growth interruption time, shifts the PL maximum to lower energies and reduces the overall full width at half-maximum (FWHM) of InAs/GaAs QDs.6 Interestingly, they also observed a distinct multimodal decomposition of the PL of InAs/GaAs QDs grown with antimony as surfactant and attributed this to shell-like ML steps of the QD size (multimodal QDs). The formation of ML steps of the QD suggests In/Ga interdiffusion is negligible under the growth condition. For sample A the Sb beam flux increases from 0 (reference sample) to 2.4 × 10-5 Pa and the PL peak is blue-shifted from 1130 to 1088 nm. This is attributed to the increased strain in the QDs as Sb incorporates into the InAs/GaAs QDs, as the lattice mismatch between InAsSb and GaAs is greater than that of InAs/GaAs QDs. From this point, the Sb surfactant effect can be excluded. This is also confirmed by our AFM results. With the incorporation of Sb, the critical thickness is reduced and the QD’s size is decreased. We believe that under our growth condition, Sb tends to incorporate to substitute for As in the group V sublattice. Most importantly, with increased excitation power, the PL wavelength from the sample A remains at the same position as shown in Figure 2b, which indicates that the band alignment of InAsSb/GaAs QDs is still type I. It is interesting to note that with the increase of the excitation power, the PL signal from excited states disappears. This is attributed to the fact that the conduction band of InAsSb is lifted up as compared to that of InAs QDs due to both the incorporation of Sb and consequent strain effect. Hence, the potential well for the electrons is shallower, resulting in a lack of any confined excited states levels inside the QDs. This is also supported by the band diagram simulations, which will be presented and discussed later on. The PL spectra of sample C in Figure 2c reveal a significant blue shift as the excitation power increases, which is a typical characteristic of type II band alignment.25-27 This blue shift is caused by Coulomb interaction between the spatially separated holes and electrons, which induces band bending and in turn affects the carrier wave functions. As the hole wave function expands, its energy level decreases. Owing to the smaller change in the electron energy level relative to the hole energy level, a blue-shifted spectrum results. The PL spectra of sample D in Figure 2d show similar blue shift of the emission with increasing the excitation power, confirming the type II band alignment of this sample. The PL measurements show markedly different behaviors for the set of investigated samples. When the Sb beam flux reaches 8 × 10-5 Pa under the growth conditions described above, clear band alignment transition from type I to type II is observed. To clarify the different optical properties of the investigated structures, the PL emission wavelength as a function of the pump power of each sample is shown in Figure 3. For the InAs QD reference sample and for sample A, clear PL signal can be still detected when the pump power © 2010 American Chemical Society
FIGURE 3. Emission wavelength as a function of the pump power for reference sample, sample A, sample C, and sample D. The size of the laser beam is around 100 µm.
is decreased down to 0.01 mW, while the PL efficiency of sample C and sample D decreases drastically, since we do not observe any optical response until the pump power reaches 0.2 and 1.23 mW, respectively. This also confirms the transition from type I to type II band alignment, where the overlap between electron and hole wave functions is significantly reduced. Another signature of type II band alignment QDs is the evolution of the carrier lifetime under different carrier densities, which differs from that observed in conventional type I InAs/GaAs QDs where each energy level experiences a single decay time scale.28,29 For a type II QD structure, a time-dependent recombination rate of nonequilibrium carriers can be observed. As a consequence, the emission dynamics at shorter wavelengths is faster in comparison to the red part of the spectrum. This is a result of dynamic Coulomb interactions between the carriers according to the mechanism explained in ref 28. To further verify the type I to type II transition of InAsSb QDs with different Sb content, TRPL measurements were performed at 8 K employing a 780 nm, 300 fs, 75.6 MHz mode-locked Ti:sapphire laser pulses as excitation source and a Hamamatsu streak camera as detection system. The decay traces in Figure 4 are extracted from the TRPL spectra of samples A, C, and D. Sample A (Figure 4a) shows no wavelength-dependent lifetimes, which confirms that it is of type I band alignment, as shown schematically in the inset. A similar situation exists for sample B. The behavior changes for sample C (Figure 4b) where a small dependence of the decay lifetime on the wavelength is observed. The red part of the spectrum decays more slowly (with a time scale of 0.8 ns) than the blue part (0.62 ns), which is indicative of a weak type II alignment. Because of this we argue that this structure is most probably close to the transition point between types I and II where the conduction band potential well is too shallow to provide confined electron states. Sample D conversely, depicted in Figure 4c, shows a profound increase of the radiative lifetimes at longer wavelengths. Additionally, the decay traces do not follow a single3054
DOI: 10.1021/nl102237n | Nano Lett. 2010, 10, 3052-–3056
FIGURE 5. Computed band profiles of InAsSb/GaAs QDs with different Sb content.
valence band of InAsSb are lifted up, leading to a decreased conduction band offset and increment of valence band offset. As Sb composition reaches around 50%, the potential does not create confined states for electrons anymore, whereas, when Sb composition reaches 70%, a type II band alignment is clearly formed. From the calculations, for Sb composition around 60% the conduction band of the InAsSb dot is at the same level as that of the GaAs barrier, achieving a flat conduction band. Moreover, comparing these calculations with our PL and TRPL results, under the growth condition described above, Sb content can be controlled in InAsSb/GaAs QDs and the conduction band offset can be systemically tuned from 0 to 500 meV, which can result in a change of the band alignment. In conclusion, we demonstrated the formation of InAsSb/ GaAs QDs. The appropriate growth conditions are described in depth. As Sb beam flux increases for different samples, power-dependent PL studies show a clear blue shift in the peak emission wavelength with increasing carrier density. Also, time-resolved photoluminescence studies show a nonsingle exponential decay indicating the transition from type I to type II band alignment. Computed band profiles show that InAsSb/GaAs QDs with 60% of Sb have a flat conduction band structure. Our study provides insights into the band engineering of such materials and will lead to the development of improved devices, such as high-efficiency solar cell and thermoelectric devices. The InAsSb/GaAs QDs are selfassembled and easily integrated into any MBE grown heterostructures and as such are more amenable to efficient electrical carrier injection and band gap engineering.
FIGURE 4. Decay traces extracted from TRPL measurement for (a) sample A, (b) sample C, and (c) sample D.
exponential behavior. Initially after the laser pulse, when the wave function overlap is significantly increased due to Coulomb interactions, the decay times are short. However the radiative events lead to continuous wave function relaxation, which results in a decreasing of the optical matrix element. As a consequence the lifetimes increase continuously, which is an indication of the type II band alignment, as schematically depicted in the inset in Figure 4c. Therefore, based on our PL and TRPL measurement results, we successfully demonstrate that type I to type II transition of InAsSb/GaAs QDs can be controlled by adjusting Sb content. In order to understand the PL and TRPL spectra and to determine the theoretical transition point between type I and type II band alignment, we calculate InAsSb/GaAs QD band structure using an eight-band, strain-dependent k·p model based on the detailed structural information obtained from the AFM. The dot shape is taken as a truncated pyramid with a height of 8 nm and base dimensions of 20 nm. The material parameters for the calculations are taken from the review article by Vurgaftman, Meyer, and Ram-Mohan30 and evaluated at 77 K. To calculate the band profile, we also need to determine the strain corrections. Toward that goal we utilize a Fourier-transform method, described in ref 31, which provides analytical solutions of the strain coefficients for the given dot shape. On the basis of these, the band profiles of InAsSb/GaAs QDs with different Sb content of 0%, 30%, 60%, and 70% are computed as shown in Figure 5. The comparison of these calculations with the PL and TRPL spectra confirm that the band structure of InAsSb/GaAs QD can be tailored by adjusting the Sb contents. From Figure 5, with the increasing of Sb content, both conduction band and © 2010 American Chemical Society
Acknowledgment. The authors gratefully acknowledge the financial support of DoD through (NSSEFF N00244-091-0091). This work was partially supported by the INSPIRE programme (funded under the Irish Government’s HEA PRTLI Cycle 4 programme and National Deveolpment Plan 2007-2013), European Commission under Marie Curie Actions (EIF Project “NextDot”, 041985) and Science Foundation Ireland (06/RFP/EWE014 and 07/IN.1/I929). 3055
DOI: 10.1021/nl102237n | Nano Lett. 2010, 10, 3052-–3056
REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Bayer, M.; Hawrylak, P.; Hinzer, K.; Fafard, S.; Korkusinski, M.; Wasilewski, Z. R.; Stern, O.; Forchel, A. Science. 2001, 291, 45. He, J.; Krenner, H. J.; Pryor, C.; Zhang, J. P.; Wu, Y.; Allen, D. G.; Morris, C. M.; Sherwin, M. S.; Petroff, P. M. Nano Lett. 2007, 7 (3), 802. Eisele, H.; Lenz, A.; Heitz, R.; Timm, R.; Da¨hne, M.; Temko, Y.; Suzuki, T.; Jacobi, K. J. Appl. Phys. 2008, 104, 124301. Marquez, H.; Geelhaar, L.; Jacob, K. Appl. Phys. Lett. 2001, 78, 2309. Eisele, H.; Flebbe, O.; Kalka, T.; Preinesberger, C.; Heinrichsdorff, F.; Krost, A.; Bimberg, D.; Dahne-Prietsch, M. Appl. Phys. Lett. 1999, 75, 106. Heitz, R.; Guffarth, R.; Potschke, K.; Schliwa, A.; Bimberg, D.; Zakharov, N. D.; Werner, P. Phys. Rev. B 2005, 71, 045325. Toda, Y.; Moriwaki, O.; Nishioka, M.; Arakawa, Y. Phys. Rev. Lett. 1999, 82, 4114. Garcia, J. M.; MedeirosRibeiro, G.; Schmidt, K.; Ngo, T.; Feng, J. L.; Lorke, A.; Kotthaus, J.; Petroff, P. M. Appl. Phys. Lett. 1997, 71, 2014. Suraprapapich, S.; Panyakeow, S.; Tu, C. W. Appl. Phys. Lett. 2007, 90, 183112. Li, S. S.; Xia, J. B. Appl. Phys. Lett. 2005, 87, No. 043102. He, J.; Notzel, R.; Offermans, P.; Koenrad, P. M.; Gong, Q.; Hamhuis, G. J.; Eijkemans, T. J.; Wolter, J. H. Appl. Phys. Lett. 2004, 85, 2771. Nishi, K.; Saito, Sugou, H. S. Appl. Phys. Lett. 1999, 74, 1111. Huffaker, D. L.; Park, G.; Zou, Z.; Shchekin, O. B.; Deppe, D. G. Appl. Phys. Lett. 1998, 73, 2564. Raghavan, S.; Rotella, P.; Stintz, A. B.; Fuchs, Krishna, S.; Morath, C.; Cardimona, D. A.; Kennerly, S. W. Appl. Phys. Lett. 2002, 91, 1369. Lu, X. J.; Vaillancourt, J.; Meisner, M. J.; Stintz, A. J. Phys. D: Appl. Phys. 2007, 40, 5878. Tsatsul’nikov, A. F.; Ledentsov, N. N.; Maksimov, M. V.; Mel’tser, B. Y.; Volovik, B. V.; Krestnikov, I. L.; Sakharov, A.; Bert, N. A.;
© 2010 American Chemical Society
(17) (18)
(19) (20) (21) (22) (23) (24) (25)
(26) (27)
(28)
(29) (30) (31)
3056
Kop’ev, P. S.; Alferov, Zh. I.; Bimberg, D. Semiconductor 1997, 31, 55. Ueta, A.; Gozu, S.; Akahane, K.; Yamamota, N.; Ohtani, N. Jpn. J. Appl. Phys. 2005, 44, No. L696. Weinhardt, L.; Ba¨r, M.; Pookpanratana, S.; Morkel, M.; Niesen, T. P.; Karg, F.; Ramanathan, K.; Contreras, M. A.; Noufi, R.; Umbach, E.; Heske, C. Appl. Phys. Lett. 2010, 96, 182102. Morkel, M.; Weinhardt, L.; Lohmu¨ller, B.; Heske, C.; Umbach, E.; Riedl, W.; Zweigart, S.; Karg, F. Appl. Phys. Lett. 2001, 79, 4484. Liu, H. F.; Xiang, N.; Chua, S. J. Appl. Phys. Lett. 2006, 89, 71905. Chen, Y. H.; Ye, X. L.; Wang, Z. G. Nanoscale Res. Lett. 2006, 1, 32. Steinshnider, J.; Harper, J.; Weimer, M.; Lin, C. H.; Pei, S. S.; Chow, D. H. Phys. Rev. Lett. 2000, 85, 4562. Pitts, O. J.; Watkins, S. P.; Wang, C. X.; Fink, V.; Kavanagh, K. L. J. Cryst. Growth 1995, 150, 858. Marquez, H.; Geelhaar, L.; Jacob, K. Appl. Phys. Lett. 2001, 78, 2309. Tatebayashi, J.; Liang, B. L.; Laghumavarapu, R. B.; Bussian, D. A.; Htoon, H.; Klimov, V.; Balakrishnan, G.; Dawson, L. R.; Huffaker, D. L. Nanotechnology 2008, 19, 295704. Kawazu, T.; Mano, T.; Noda, T.; Sakaki, H. Appl. Phys. Lett. 2009, 94, No. 081911. Hayne, M.; Maes, J.; Bersier, S.; Moshchalkov, V. V.; Schliwa, A.; Muller-Kirsch, L.; Kapteyn, C.; Heitz, R.; Bimberg, D. Appl. Phys. Lett. 2003, 82, 4355. Gradkowski, K.; Pavarelli, N.; Ochalski, T. J.; Williams, D. P.; Tatebayashi, J.; Huyet, G. P.; O’Reilly, E.; Huffaker, D. L. Appl. Phys. Lett. 2009, 95, 061102. Liao, Y. A.; Hsu, W. T.; Lee, M. C.; Chiu, P. C.; Chyi, J. I.; Chang, W. H. Phys. Status Solidi 2009, C 6, 1449. Vurgaftman, I. J.; Meyer, R.; Ram-Mohan, L. R. J. Appl. Phys. 2001, 89, 5815. Andreev, A. D.; Downes, J. R.; Faux, D. A.; O’Reilly, E. P. J. Appl. Phys. 1999, 86, 297.
DOI: 10.1021/nl102237n | Nano Lett. 2010, 10, 3052-–3056
