Single Excitons in InGaN Quantum Dots on GaN Pyramid Arrays

From μPL, it is evident that the QDs are located in the apexes of the pyramids. The fact that the emission lines of the QDs are linear polarized in a...
0 downloads 0 Views 2MB Size
LETTER pubs.acs.org/NanoLett

Single Excitons in InGaN Quantum Dots on GaN Pyramid Arrays Chih-Wei Hsu,† Anders Lundskog,† K. Fredrik Karlsson, Urban Forsberg, Erik Janzen, and Per Olof Holtz* Department of Physics, Chemistry and Biology (IFM), Link€oping University, S-581 83, Link€oping, Sweden ABSTRACT: Fabrication of single InGaN quantum dots (QDs) on top of GaN micropyramids is reported. The formation of single QDs is evidenced by showing single sub-millielectronvolt emission lines in microphotoluminescence (μPL) spectra. Tunable QD emission energy by varying the growth temperature of the InGaN layers is also demonstrated. From μPL, it is evident that the QDs are located in the apexes of the pyramids. The fact that the emission lines of the QDs are linear polarized in a preferred direction implies that the apexes induce unidirected anisotropic fields to the QDs. The single emission lines remain unchanged with increasing the excitation power and/or crystal temperature. An in-plane elongated QD forming a shallow potential with an equal number of trapped electrons and holes is proposed to explain the absence of other exciton complexes. KEYWORDS: InGaN, quantum dots, pyramid, exciton, photoluminescence

S

emiconductor quantum dots (QDs) have been demonstrated to play an important role and will play an even more important role in various light-emitting devices. For example, improved efficiencies of the light-emitting diodes and laser diodes can be achieved by incorporating QDs in the active layer.1 The photons emitted from QDs possess specific energies and could be timecorrelated and/or quantum-entangled. Such single photon characteristics are promising for quantum cryptographic (QCA) and other quantum information applications (QIA).24 In these applications, one QD represents an individual quantum light source and it is of utmost importance that the QD can be positioned in a controllable manner in order to facilitate the subsequent device processing.5 Owing to the lack of high-quality substrate, the progress of the nitride-based QDs emitting in the UV/blue regime is left behind the corresponding arsenide-based QDs, emitting in the near-IR regime. Several important features for QCA and QIA, like polarization-entanglement and single-photon emission in a sitecontrolled fashion, have only been demonstrated for arsenidebased QDs.6,7 However, all arsenide based devices for QCA and QIA face the drawback of being operating just at cryogenic temperatures. InGaN/GaN QDs allow deeper confinement potential with the possible advantage for high temperature operation.8 In addition, devices working at shorter wavelengths enable reduced size of the optical interconnect as well as full-color generation in the visible spectrum. In the literature, most of the InGaN QDs are grown self-assembled on planar substrates in the Stranski Krastanow (SK) growth mode. However, the SK-grown InGaN QDs often suffer from random growth position and inhomogeneous size,911 which greatly raise the difficulty of utilizing these QDs for frontier line quantum light emitter applications. Detailed investigations on individual InGaN QD rely on postmasking process in order to minimize the number of QDs exposed to the excitation.12 For such a self-assembled approach, reducing the dimension of the hosting template from a two-dimensional r 2011 American Chemical Society

layer to one-dimensional nanowires could be one solution to this issue. However, attempts made by a molecular beam epitaxy13 (MBE) was found to give better result than a conventional chemical vapor deposition (CVD) system.14 Up to date, the pyramidal-structured templates, which provide preferential formation locations for the QDs, are the most common way of achieving site control of the InGaN QDs.1517 However, reports on InGaN QDs grown on pyramids showing well-resolved emission peaks are rare.17 The sharp emission is a direct consequence of threedimensional quantum confinement with fully quantized energy levels of QDs. The limited full width at half-maximum (fwhm) of the emission is particularly important for QIA.18 In this Letter, sharp emission lines (sub-millielectronvolts fwhm) from single site-controlled InGaN QDs located at the apexes of hexagonal GaN pyramids is reported. The emission energies of the QDs are demonstrated to be tunable within a certain energy range by varying the growth temperatures of the active InGaN layer. In addition, the emission lines of single QDs on various pyramids tend to be polarized in the Æ1100æ direction as concluded from microphotoluminescence (μPL) measurements. The ability to fabricate single InGaN QDs in a controllable fashion shows the potential for these single QDs as a quantum light emitter (QLE). The complete structure is grown by a horizontal low pressure hot-wall metalorganic CVD using trimethylgallium, trimethylindium, and NH3 as precursors. A ∼100 nm AlN nucleation layer, followed by a 2 μm GaN layer, is deposited on a Si-face onaxis 4H-SiC substrate.19 Selective area growth of the GaN pyramids is achieved by regrowth of the GaN on the lithographically patterned SiN-masked GaN/AlN/SiC template; circular openings with the diameter of 2.5 μm forming square blocks with 25  25 Received: March 10, 2011 Revised: April 15, 2011 Published: April 28, 2011 2415

dx.doi.org/10.1021/nl200810v | Nano Lett. 2011, 11, 2415–2418

Nano Letters

Figure 1. (a) A tilted SEM image of the InGaN QDs grown on GaN pyramids. Schematic illustrations of possible QD formation mechanisms are shown in (b) and (c). (b) Dot formation due to thickness enhancement and indium enrichments of the InGaN layer at the apex. (c) Dot formation due to the coherent SK-growth mode at the apex.

openings are repeated over the template. After growth of the GaN pyramids, an InGaN layer followed by a GaN capping layer is deposited on the pyramids at temperatures between 710 and 750 C. The pyramid pitch was varied from 4.1 to 6.5 μm across the sample, resulting in a pyramid density ranging from 2.4 to 5.9  106 cm2. A μPL setup with accessible temperatures of 300 down to 4 K was employed to investigate the optical properties of the InGaN QDs. A continuous laser working at 266 nm with a maximal accessible power of 6 mW (measured right before objective) was used as the excitation source. A focused spot diameter of ∼1.5 μm can be reached by a reflective objective. The emission is collected by the same objective and subsequently guided to a monochromator (50 cm focal length) with a 2400 grooves/mm grating and a spectral resolution of about 350 μeV in the studied energy regime. A representative scanning electron microscopy image of the complete structure reveals that the edges of the pyramids are 1.5 μm in width and 3.2 μm in height (Figure 1a). The pyramids are grown in the Æ0001æ direction with six equivalent facets of {1011} owing to the hexagonal wurtzite crystal symmetry of GaN. The pyramid pitch is significantly larger than the laser spot, allowing us to exclude the interference from neighboring pyramids. The maximum PL intensity of sharp emission peaks is achieved when the excitation is focused directly on the apex of the pyramid. The intensity decreases as the laser spot is shifted away from the apex and disappears at a distance of approximately 2 μm from the apex, confirming that the QDs are preferably formed on the pyramid apexes. In practice, examining the thickness and the local In composition at the pyramid apex is difficult. Reports on InGaN QWs grown on top of pyramidal GaN stripes revealed that the InGaN grew faster on the [0001] plane compared to the {1101} planes.20 Additionally, In also tends to accumulate at the stripe ridge, causing compositional fluctuations.20 Both thickness and compositional fluctuations are known to generate QD-like localization centers in InGaN MQW structures.2024 Thus, the formation of QDs can be the result of (1) thickness enhancement20,25 and indium enrichments23 of the InGaN layer at the apex of the hexagonal GaN pyramid as a consequence of the enhanced source

LETTER

Figure 2. (a) Representative μPL spectra from pyramids hosting single QDs grown at different temperatures. (b) A polar plot of the normalized PL intensity against the recorded polarization direction (θ). The polarization direction parallel to the Æ1100æ of the GaN crystal is set to θ = 0. (c) The polarization directions and ratios for the corresponding QDs shown in (a).

supply (Figure 1b) or (2) SK-like growth mode with transitions from layer by layer growth to island growth on a microscopic c-plane area on top of the pyramid (Figure 1c). In alternative (1), the QDs are expected to exhibit a red shift of the emission energy with respect to their surrounding QWs. Both thicker InGaN layer at the pyramidal apex and higher In concentration tend to lower the emission energy, as reported in ref 17 and similar structure of GaAs QDs formed in inverted pyramids.26 On the other hand, QDs formed via alternative (2) could be isolated from the QWs grown on the side-wall of the pyramids, leading to higher emission energy of the QDs. In addition, we find that the QD formation probability is significantly increased on pyramids with a notably small c-plane truncated apex, supporting alternative (2) as the favorable formation mechanism.27 Figure 2a shows several μPL spectra measured on pyramids hosting single QDs at 4 K. All spectra are dominated by single emission peaks in the wavelength range of 390460 nm. The emission energy of the QDs could be tuned by changing the growth temperature of the active InGaN layer from 750 to 710 C. The observed red shift with decreasing growth temperature is believed to be due to the enhanced indium incorporation at lower temperature. The typical observed fwhm of the emission lines is in the range of 3501200 μeV, which is comparable to reported values for QDs grown on planar23,28 and pyramidal substrates.29 The fwhm could be increased due to the effect of spectral diffusion,30 but it should also be noted that the minimum observed fwhm of 350 μeV could be limited by the resolution of the monochromator. Surprisingly, the emission lines from various QDs are linearly polarized with a typical polarization ratio >0.931 (Figure 2b) and exhibit a preferred polarization direction along the Æ1100æ direction (Figure 2c). Since the linear polarization and the emission intensity of the QDs are independent of the linear polarization direction of the excitation laser, the observed linear polarization indicates an in-plane anisotropy of the QD potential itself.32,33 The preferential polarization direction observed here is unlike other studied InGaN and GaN QDs grown by molecular beam epitaxy, for which no preferred polarization ratios and directions are observed for the individual QDs.34,35 The pyramidal template seems to provide a reproducible anisotropy 2416

dx.doi.org/10.1021/nl200810v |Nano Lett. 2011, 11, 2415–2418

Nano Letters

Figure 3. Power-dependent μPL measurements of a selected QD performed at 4 K. Inset: A corresponding logarithmic plot of the μPL integrated intensity as a function of the applied excitation power. The solid red line represents the slope = 1.

LETTER

pyramid. The peak energy of X experiences a red shift of 1.5 meV by increasing the temperature from 4 to 70 K, similar to the reports of other InGaN QDs.9,28 Meanwhile, the fwhm of the X increases from 700 μeV (4 K) to 1.1 meV (70 K). The slowly increasing fwhm with respect to the temperature could be dominated by the excitonacoustic phonon interaction37 and is suggested to be a characteristic feature of a QD structure.9,28,37 The demonstrated temperature insensitive nature of the nitride QDs is essential for high temperature operation. The fully quantized energy level scheme for the electron (e) and the hole (h) in a QD is associated with distinctly different emission energies corresponding to a unique QD charge population. Commonly observed emission lines are attributed to the singleexciton (1e1h), charged-exciton (1e2h or 2e1h), biexciton (2e2h). The observation of only the single-exciton implies that the QD constitutes a relatively shallow potential, for which any other exciton complexes are unbound. The computed repulsive ee (hh) Coulomb energies are about 10 (15) meV larger than the attractive eh energy for small QDs (2 nm height).38 The difference between the repulsive and attractive energies can become even more prominent by increasing the QD size.38 Thus, all other exciton complexes involving more than one e and h could be energetically unfavorable owing to the large repulsive Coulomb forces between the ee and/or hh, explaining their absence in the μPL spectra. In conclusion, well-defined and sub-millielectronvolt emission peaks originating from single and neutral excitons from InGaN QDs on top of GaN pyramids are reported. The linearly polarized emission lines with a preferred polarization direction along Æ1100æ direction are explained in terms of systematic pyramidinduced anisotropic fields. Power- and temperature-dependent μPL measurements suggest that the QDs studied are bound in inplane elongated and relatively shallow potentials. Our results reveal a promising concept for generation of single photons from pyramidal InGaN/GaN QDs.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Figure 4. Temperature-dependent μPL measurements of a selected QD (the same QD as shown in Figure 3). Inset: A plot of the peak energies as a function of the temperature.

of the QD potential, which could be a great advantage for QLE requiring polarization control. On the other hand, such anisotropy prohibits the emission of polarization entangled photon pairs. For a selected QD measured at 4 K, only one emission peak, centered at 2.9705 eV, with a fwhm of 700 μeV is observed (shown in Figure 3). The intensity of X grows linearly with the excitation power (Figure 3, inset), suggesting the identification as the single exciton emission. No significant energy shift or appearance of additional emission lines is observed in the excitation power range of 45 μW to 6 mW (Figure 3). Such a PL behavior, with independence on the excitation power, is different from the reported results for similar structures.29 Moreover, the single peak remains single, also when the temperature is increased from 4 to 70 K (Figure 4). The absence of any temperature-induced redistribution, as observed for QD ensembles at moderate temperatures,36 further supports our interpretation that the emission originates from an individual QD on the

Author Contributions †

C. W. Hsu and A. Lundskog contributed equally to this work.

’ ACKNOWLEDGMENT This work was supported by the NANO-N consortium funded by the Swedish Foundation for Strategic Research (SSF). ’ REFERENCES (1) Knill, E.; Laflamme, R.; Milbum, G. J. Nature 2001, 46, 409. (2) Gisin, N.; Ribordy, G.; Tittel, W.; Zbinden, H. Rev. Mod. Phys. 2002, 74, 145. (3) De Rinaldis, S.; D’Amico, I.; Rossi, F. Phys. Rev. B 2004, 69, 235316. (4) Bouwmeester, D., Ekert, A. K., Zeilinger, A. The Physics of Quantum Information; Springer: Berlin, 2000. (5) Gaisler, V. A. Bull. Russ. Acad. Sci.: Phys. 2009, 73, 77–79. (6) Mohan, A.; Felici, M.; Gallo, P.; Dwir, B.; Rudra, A.; Faist, J.; Kapon, E. Nat. Photonics 2010, 4, 302–306. (7) Schneider, C.; Heindel, T.; Huggenberger, A.; Weinmann, P.; Kistner, C.; Kamp, M.; Reitzenstein, S.; H€ofling, S.; Forchel, A. Appl. Phys. Lett. 2009, 94, 111111. 2417

dx.doi.org/10.1021/nl200810v |Nano Lett. 2011, 11, 2415–2418

Nano Letters (8) Kako, S.; Santori, C.; Hoshino, K.; Gotzinger, S.; Yamamoto, Y.; Arakawa, Y. Nat. Mater. 2006, 5, 887. (9) Moriwaki, O.; Someya, T.; Tachibana, K.; Ishida, S. Appl. Phys. Lett. 2000, 76, 2361–2363. (10) Bayram, C.; Razeghi, M. Appl. Phys. A: Mater. Sci. Process. 2009, 96, 403–408. (11) Davies, S. C.; Mowbray, D. J.; Ranalli, F.; Parbrook, P. J.; Wang, Q.; Wang, T.; Yea, B. S.; Sherliker, B. J.; Halsall, M. P.; Kashtiban, R. J.; Bangert, U. Appl. Phys. Lett. 2009, 95, 111903. (12) Sch€omig, H.; Halm, S.; Forchel, A.; Bacher, G.; Off, J.; Scholz, F. Phys. Rev. Lett. 2004, 92, 106802. (13) Renard, J.; Songmuang, R.; Bougerol, C.; Daudin, B.; Gayrol, B. Nano Lett. 2008, 7, 2092–2096. (14) Hsu, C. W.; Ganguly, A.; Liang, C. H.; Hung, Y. T.; Wu, C. T.; Hsu, G. M.; Chen, Y. F.; Chen, C. C.; Chen, K. H.; Chen, L. C. Adv. Funct. Mater. 2008, 18, 938–942. (15) Tachibana, K.; Someya, T.; Ishida, S.; Arakawa, Y. Appl. Phys. Lett. 2000, 76, 3212–3214. (16) Perez-Solorzano, V.; Gr€oning, A.; Jetter, M.; Riemann, T.; Christen., J. Appl. Phys. Lett. 2005, 87, 163121. (17) Edwards, P. R.; Martin, R. W.; Watson, I. M.; Liu, C.; Taylor, R. A.; Rice, J. H.; Na, J. H.; Robinson, J. W.; Smith, J. D. Appl. Phys. Lett. 2004, 85, 4281. (18) Jarjour, A. F.; Oliver, R. A.; Taylor, R. A. Philos. Mag. 2007, 87, 2077–2093. (19) Forsberg, U.; Lundskog, A.; Kakanakova-Georgieva, A.; Ciechonski, R.; Janzen, E. J. Cryst. Growth 2009, 311, 3007–3010. (20) Nishizuka, K.; Funato, M.; Kawakami, Y.; Fujta, S.; Narukawa, Y.; Mukai, T. Appl. Phys. Lett. 2004, 85, 3122–3124. (21) Feng, W.; Kuryatkov, V. V.; Chandolu, A.; Song, D. Y.; Pandikunta, M.; Nikishin, S. A.; Holtz, M. J. Appl. Phys. 2008, 104, 103530. (22) Yoshikawa, M.; Murakami, M.; Ishida, H.; Harima, H. Appl. Phys. Lett. 2009, 94, 131908. (23) Seguin, R.; Rodt, S.; Strittmatterm, A.; Reissmann, L.; Bartel, T.; Hoffmann, A.; Bimberg, D.; Hahn, E.; Gerthsen, D. Appl. Phys. Lett. 2004, 84, 4023–4025. (24) Chichibu, S. F.; Uedono, A.; Onuma, T.; Haskell, B. A.; Chakraborty, A.; Koyama, T.; Fini, P. T.; Keller, S.; Denbaars, S. P.; Speck, J. S.; Mishra, U. K.; Nakamura, S.; Yamaguchi, S.; Kariyama, S.; Amano, H.; Akasaki, I.; Han, J.; Sota, T. Philos. Mag. 2007, 87, 2019–2039. (25) Biasiol, G.; Kapon, E. Phys. Rev. Lett. 1998, 81, 2962–2965. (26) Hartmann, A.; Ducommun, Y.; Leifer, K.; Kapon, E. J, Phys.: Condens. Matter 1999, 11, 5901–5915. (27) Lundskog, A., Hsu, C. W., Karlsson, K. F., Forsberg, U., Holtz, P. O., Janzen, E. In preparation. (28) Oliver, R. A.; Briggs, G. A. D.; Kappers, M. J.; Humphreys, C. J.; Yasin, S.; Rice, J. H.; Smith, J. D.; Taylor, R. A. Appl. Phys. Lett. 2003, 83, 755–757. (29) Martin, R. W.; Edwards, P. R.; Taylor, R. A.; Rice, J. H.; Na, J. H.; Robinson, J. W.; Smith, J. D.; Liu, C.; Watson, I. M. Phys. Status Solidi A 2005, 202, 372–376. (30) Empedocles, S. A.; Norris, D. J.; Bawendi, M. D. Phys. Rev. Lett. 1996, 77, 3873. (31) The polarization ratio is defined as (Imax  Imin)/(Imax þ Imin). The Imax is the maximum PL intensity recorded at a certain polarization direction, θ, depending on the experimental result. The Imin is the PL intensity recorded at the polarization direction perpendicular to θ, which is θ þ 90. (32) Winkelnkemper, M.; Seguin, R.; Rodt, S.; Schliwa, A.; Reiβmann, L.; Strittmatter, A.; Hoffmann, A.; Bimberg, D. J. Appl. Phys. 2007, 101, 113708. (33) Sugisaki, M.; Ren, H. W.; Nair, S. V.; Nishi, K.; Sugou, S.; Okuno, T.; Masumoto, Y. Phys. Rev. B 1999, 59, R5300–R5303. (34) Supaluck, A.; Chen, Y. T.; Karlsson, K. F.; Chen, K. H.; Hsu, H, C.; Hsiao, C. L.; Chen, L. C.; Holtz, P. O. Phys. Rev. B Accepted. (35) Bardoux, B.; Guillet, T.; Gil, B.; Levebvre, P.; Bretagnon, T.; Taliercio, T.; Rousset, S.; Semond, F. Phys. Rev. B 2008, 77, 235315.

LETTER

(36) Karlsson, F. K.; Moskalenko, E. S.; Holtz, P. O.; Monemar, B.; Schoenfeld, W. V.; Garcia, J. M.; Petroff, P. M. Appl. Phys. Lett. 2001, 78, 2952–2954. (37) Asaoka, K.; Ohna, Y.; Kishimoto, S.; Mizutani, T. Jpn. J. Appl. Phys. 1999, 38, 546–549. (38) Willams, D. P.; Andreev, A. D.; O’Reilly, E. P. Superlattices Microstruct. 2004, 36, 791–798.

2418

dx.doi.org/10.1021/nl200810v |Nano Lett. 2011, 11, 2415–2418