Optically Bright Quantum Dots in Single Nanowires - Nano Letters

Bright Single InAsP Quantum Dots at Telecom Wavelengths in Position-Controlled InP Nanowires: The Role of the Photonic Waveguide. Sofiane HaffouzKatha...
0 downloads 5 Views 258KB Size
NANO LETTERS

Optically Bright Quantum Dots in Single Nanowires

2005 Vol. 5, No. 7 1439-1443

Magnus T. Borgstro1 m,*,† Valery Zwiller,† Elisabeth Mu1 ller,‡ and Atac Imamoglu† Quantum Photonics Group, Institute of Quantum Electronics, ETH Ho¨nggerberg HPT G10, CH 8093, Zu¨rich, Switzerland, and Solid State Physics, ETH Ho¨nggerberg, CH 8093, Zu¨rich, Switzerland Received April 29, 2005; Revised Manuscript Received June 2, 2005

ABSTRACT We fabricate and demonstrate optically active quantum dots embedded in single nanowires. Observation of photon antibunching proves the zero dimensionality of these heterostructures that can be epitaxially grown on various substrates, including silicon. We show that the nanowire dots are intense single photon sources, typically an order of magnitude brighter than self-assembled quantum dots. Due to control over their composition, size, and position, nanowire dots are ideal building blocks for fully controlled quantum dot molecules.

Optically active quantum dots (QDs) are at the center-stage of nanoscience due to their novel physical properties as well as potential applications ranging from biological marking to single-photon sources.1,2 Investigation of physical phenomena and applications requiring three or more QDs have, however, been severely hindered by the limited control one has over the growth process. In parallel to the QD research, several groups have studied electronic and optical properties of heterostructure nanowires.3-9 Here we report the first experimental observation of optically bright QDs in a single nanowire where confinement length-scales are externally controllable. We demonstrate three different optically active QDs in different sections of a nanowire, showing the potential of this technique in realizing control over the size and material composition of, as well as the distance between, neighboring QDs. Despite the proximity of confined excitons to the surface in a 20 nm diameter wire, the QDs reported are more than an order of magnitude brighter than selfassembled InAs dots and exhibit strong photon antibunching, proving the zero-dimensional nature of optical excitations. Low-pressure metal organic vapor phase epitaxy (LPMOVPE) was used for nanowire growth, following previous experiments where metal catalyst particles have been used to control the diameter10-12 (defined by catalyst size), length (set by growth time),13 position,14 and composition4,5,15 of semiconductor nanowires by a vapor-liquid-solid (VLS)16 growth process. 20- and 40 nm colloidal Au solutions were allowed to interact with lysine-coated, pre-polished Si (001)/ HF (room temperature) and 〈111〉B GaP/HCl/H2O/HNO3 (3: 3:2, 80 °C), substrates for 5 and 20 s respectively, after which the samples were blown dry with N2. The samples were then * Corresponding author. E-mail: [email protected]. † Institute of Quantum Electronics. ‡ Solid State Physics. 10.1021/nl050802y CCC: $30.25 Published on Web 06/16/2005

© 2005 American Chemical Society

transferred to a LP-MOVPE system (12 l/min H2 at 100 mbar) and placed on a gas foil rotated graphite disk on a graphite susceptor, heated by halogen lamps. The samples were preheated at 580 °C for a total of 17 min to desorb any surface oxide and alloy the Au with GaP, primarily by an up-take of Ga into the Au droplet, before cooling down to 440 °C during a 5 min period. Growth of nanowires was then initiated by introducing trimethyl-gallium (TMG) into the reactor cell at a partial pressure of 1.2 × 10-3 mbar. The compositions in the wires were controlled by adjusting the arsine-to-phosphine ratio, keeping the phosphine and TMG flows constant, and 15 nm QD segments were embedded in either 200- or 500 nm long direct band gap (GaAs1-xPx, x < 0.45) segments in the indirect band gap GaP nanowire that absorb excitation laser and guide excited carriers to the QDs, free from dislocations due to efficient strain relaxation in nanowires where the conventional requirement for lattice matching is circumvented. No luminescence could be detected from wires excited by incoming laser wavevector parallel to the wires, indicating wire polarization anisotropy in accordance with previous experiments on single InP wires.17 To avoid this problem, wires were mechanically transferred to a patterned SiO2 substrate before optical characterization. Scanning electron microscopic (SEM) images (Figure 1a) correlate with microphotoluminescence images of the corresponding region (Figure 1b) and with optical microscopy images (Figure 1c). Most of the wires studied were found either with a small angle toward the SiO2 substrate, at one of the patterns as seen in Figure 1, or lying on top of other wires. No difference in optical properties of the wires was observed when using patterns fabricated with different techniques (etched in a thermally oxidized Si wafer or defined by Au evaporation and lift off) to facilitate locating and identification of the

Figure 1. QD nanowire heterojunction. (a) SEM image of two 40 nm wires, 6 µm long. (b) Photoluminescence image of the wires depicted in (a) where bright sections correspond to QD regions. The upper and lower QDs, 3.1 µm apart and embedded in 200 nm direct band gap absorption layers, shine strongly, whereas the middle QD luminescence is weak. The structure schematic can be found in Figure 4. (c) Microscope image of the same wires under LED illumination. Arrows indicate the base of the wires. (d,e) Elemental maps of the P (blue) and As (red) content of the middle QD region in a 40 nm nanowire, obtained by EFTEM. (f) Line scans deduced from elemental maps of the QD region in a 20 nm wire verify the core/shell structure.

wires. For transmission electron microscopy, wires were transferred to copper grids and images were collected on a Philips CM 30ST (high resolution work) and a FEI Tecnai F30ST (analytical work), both being operated at 300 kV. Energy filtered transmission electron microscopy (EFTEM) measurements were consistent with energy-dispersive X-ray spectroscopy measurements, both methods being carried out for elemental analysis. In the line scans deduced from the EFTEM measurements, the intensities were averaged over 200 pixels along the wire and As scans were multiplied by a factor of 10, as compared to P, to compensate for the lower counting rate at the higher energy loss of the As-edge. The As- and P-maps of the middle QD region (Figure 1 d,e), out of three, reveal the presence of a P-rich shell enclosing an As containing core. Line scans perpendicular to the growth direction across the active region of a 20 nm wire deduced from elemental maps are shown in Figure 1f. The lateral dimension of the wire was 20 (31) nm at the top (base) of the wire respectively, determined by high-resolution transmission electron microscopy, suggesting a few nm thick tapering induced shell at half the wire length, which in turn is expected to reduce surface effects. 1440

Optical characterization was performed with a low-temperature confocal microscope: the output of a mode-locked picosecond (3 ps) Ti:sapphire laser was frequency doubled (430 nm) to excite the quantum dots. Figure 2a is a schematic of the experimental setup. The photoluminescence was sent to a spectrometer (30 µeV spectral resolution), and a streak camera (40 ps time resolution) or a Hanbury-Brown and Twiss setup (1.2 ns time resolution) was used for photon correlation measurements. To measure photon correlations, the number of pairs of photons n(τ) with arrival-time separations of τ was measured using the single-photon counting modules (APD) depicted in Figure 2a. The outputs were then fed into a multichannel analyzer. Photoluminescence spectra recorded as a function of excitation power on a single QD in a single 20 nm diameter wire (Figure 2b) show a single emission line at 1.86 eV with a full width at half-maximum (fwhm) of 2.1 meV at low excitation power. Figure 2c (lower plot) shows the results of photon correlation measurements carried out using the Hanbury-Brown and Twiss setup under pulsed laser excitation of the QD:18 since the area of the coincidence peak at τ ) 0 is 6 times smaller than that of successive peaks, we conclude that the luminescence source is necessarily a single anharmonic quantum emitter (that is, a single QD); if the luminescence (Figure 2c, middle plot) had originated from two or more QDs, the area of the peak at τ ) 0 would have been at least half that of the other peaks. This observed “photon antibunching” is a direct signature of the fact that a single QD cannot emit more than one photon per excitation at the fundamental exciton transition, ideally leading to vanishing coincidence counts in a correlation measurement. The finite value of the τ ) 0 peak in our experiment is due to the fact that we could not filter out the biexciton from the exciton luminescence completely (described below). Measurements carried out on 10 different wires all exhibited similar levels of antibunching. As expected, photon correlation measurements on the excitation laser source (depicted in Figure 2c, upper plot) exhibit no photon antibunching. Given the adverse role of proximity to surfaces in selfassembled QDs,19 one could have expected the luminescence yield of nanowire QDs to be low due to nonradiative surface recombination. In stark contrast, nanowire QDs prove to be extremely efficient single photon sources,1,2 and antibunching could be observed with more than one million photon detections per detector per second. The apparent brightness of the QDs is most likely due to the fact that they are not suffering from being embedded in a high refractive index material, unlike self-assembled QDs. Neither blinking nor bleaching was observed on these QDs: the core-shell structure observed in the TEM studies could be playing a crucial role in suppressing blinking and nonradiative surface recombination. Another expected effect of surface proximity is luminescence line width broadening due to surface states, as observed on self-assembled QDs in etched mesa structures when the diameter of the mesa was made smaller than 100 nm.20 The dots under study emit with line widths of a few hundreds of µeV up to about 3 meV, and the emission wavelength varies over several meV from dot to dot, even when grown on the same substrate and in the same run. The Nano Lett., Vol. 5, No. 7, 2005

Figure 2. Luminescent single 20 nm nanowire QD. (a) Schematic of the experimental setup. BSa: beam sampler, PH: pinhole, BPF: band-pass filter, BSp: beam splitter. (b) Power-dependent spectra on the QD (VPH3/VAsH3 ) 10) embedded in 500 nm direct band gap absorption layers recorded at 5.2 K after 10 s integration. At low power, a single exciton emission line is observed. With increasing pump power, the biexciton emission appears on the low energy side of the exciton peak. (c) Top: photon correlation measurement performed on pulsed Ti:sapphire laser emission. Bottom: Experimental demonstration of strong photon antibunching and single photon emission from the QD. The inset shows the 10 nm band-pass filtered QD-emission sent to the HBT setup.

narrowest line width observed from any GaAsP QD was about 200 µeV and was grown on Si. We also measured the dependence of QD luminescence on temperature: we observed that the QD luminescence was quenched with increasing temperature, and it vanished above T ) 77 K. A red shift of the luminescence peak (12 meV from 10 to 65 K) is in accord with the expected band gap shrinkage. In the power dependence measurements, an additional emission line appeared 2.9 meV below the exciton peak with increasing excitation power (Figure 2b). The pump power dependence of this peak strongly suggests that it originates from biexciton recombination. The ascribed exciton and biexciton intensities are seen to saturate at high pumping powers. To verify the exciton/biexciton behavior, timeresolved photoluminescence measurements were recorded under different excitation power on the dot and on several other dots from samples with either 500- or 200 nm absorption layers. When using 500 nm GaAsxP1-x (VPH3/VAsH3 ) 40) absorption layers around the QD (data in Figures 2 and 3), we observe a monoexponential decay of the PL intensity at low excitation power for the exciton peak at 1.86 eV, with a radiative lifetime of 0.42 ns (Figure 3a, b). The presence of a low-energy tail in the emission spectrum indicates that some biexciton recombination contributes to photon emission, even at the lowest excitation power. The exciton and biexciton lifetimes were measured to be 0.53 ns and 0.28 ns at the highest excitation power of 2 kWcm-2. We observe the exciton peak intensity to be delayed relative to the biexciton under strong laser excitation, demonstrating cascaded biexciton and exciton emission, with the biexciton photon being emitted before the exciton photon.21 The radiative lifetimes and strong luminescence indicate that no Nano Lett., Vol. 5, No. 7, 2005

strong nonradiative processes take place. In experiments where thin 200 nm absorption segments were used, a biexponential decay of the PL intensity as a function of time could often be observed, with time constants of 0.5 ns and 20 ns indicating a slow feeding of the QD with carriers from the indirect band gap GaP region. Only few of the wires grown on Si22 substrate showed sharp, QD-like emission (about 1/50). A higher growth rate of nanowires on Si than on GaP where some material is lost to competing substrate epitaxy, was observed by SEM (a factor ≈ 1.7, the diameter is defined by the Au particle, except for a small tapering often seen on MOVPE grown wires).13,23 Therefore, larger QDs may be expected in wires grown on Si than on GaP under otherwise identical growth conditions. Luminescence from three QDs in a single 40 nm diameter wire was measured simultaneously by defocusing the excitation laser. The spectra (Figure 4) show that the emission energy of the dots can be tuned from 660 to 700 nm by adjusting the VPH3/VAsH3 ratio during growth (QD:VPH3/VAsH3 ) 20, 10 and 0, in GaAsP:VPH3/VAsH3 ) 60 absorption layers). The P content in the QD material scales with the composition in the gas phase during growth, and the observed QD emission decreases linearly with decreasing PH3 pressures, which in a first approximation reflects the resulting band gap change, demonstrating the potential of controlling QD emission energy by growth parameters. The luminescence energy (1.77 eV) of the nominally pure GaAs (Eg ) 1.52 eV) QD is too blue-shifted to be explained by simple modeling with one dominant quantization axis, in this case the vertical direction along the wire. We cannot attribute the energy shift to a reactor memory effect leading to higher P content in the material grown since wire luminescence from 200 nm 1441

Figure 4. Tunability of QD emission. (a) Three QDs with different materials composition in a single 40 nm nanowire exhibit three emission energies. The observed intensity change of the respective emission peaks is due to unequal excitation of the different regions due to laser focal position. The distance between each QD segment was 1.2 µm, and the most intensely luminescent dot could be chosen by moving the sample position. A schematic of the nanowire is inset below the spectrum. (b) Switching procedure of chemicals during growth of the active region. A growth interrupt of three seconds was introduced before and after QD growth, to ramp the AsH3 to the flow defined for the corresponding VPH3/VAsH3 ratio.

but not least, the demonstration of optically active QDs grown on Si substrate may enable the dream of efficiently coupling electronics and optics. Acknowledgment. We thank E. Gini and M. Ebno¨ther for support on LP-MOVPE, and C. Ellenberger for metal evaporation. Furthermore, we thank C.-W. Lai for assistance with optical measurements. This work was supported by internal grants from ETH Zurich and by QSIT. Figure 3. Radiative lifetimes in a nanowire QD (a) Time-resolved measurements on the photoluminescence from the single QD presented in Figure 2 under different laser excitation intensities. (b) Line scans showing the radiative lifetime for the exciton and biexciton under 50- and 2000 Wcm-2 excitation power. The exciton emission peak shifts in time with increasing pumping power due to feeding from the biexciton decay.

long segments of the same composition showed luminescence close to that expected for GaAs. The vertical position of optically bright QDs within a nanowire was defined in-situ during growth, which, together with the possibility of laterally defining the position of nanowires as shown by Mårtensson et al.,14 opens up a wide field of new possibilities of nanoscale design for quantum optics and photonics, such as bottom-up fabrication of photonic crystals with embedded QDs. In addition, combined with the prior demonstration of monolayer control over the material composition in a nanowire,5 our findings establish a first step toward design of quantum dot molecules or chains, where coupling between charges (spins) in essentially identical QDs can be mediated via tunneling (Heisenberg24) or via dipole-dipole (optically mediated RKKY25) interactions. Such structures have been analyzed for their potential applications in quantum information processing.24-27 It should also be emphasized that a single nanowire is in many ways an ideal nanomechanical device, and the possibility of incorporating QDs in a nanowire may lead to laser cooling and manipulation of mechanical degrees of freedom.28 Last 1442

References (1) Michler, P.; Kiraz, A.; Becher, C.; Schoenfeld, W. V.; Petroff, P. M.; Zhang, L. D.; Hu, E.; Imamoglu, A. Science 2000, 290, 2282. (2) Santori, C.; Fattal, D.; Vuckovic, J.; Solomon, G. S.; Yamamoto, Y. Nature 2002, 419, 594. (3) Hiruma, K.; Murakoshi, H.; Yazawa, M.; Katsuyama, T. J. Cryst. Growth 1996, 163, 226. (4) Gudiksen, M. S.; Lauhon, L. J.; Wang, J.; Smith, D. C.; Lieber, C. M. Nature 2002, 415, 617. (5) Bjork, M. T.; Ohlsson, B. J.; Sass, T.; Persson, A. I.; Thelander, C.; Magnusson, M. H.; Deppert, K.; Wallenberg, L. R.; Samuelson, L. Appl. Phys. Lett. 2002, 80, 1058. (6) Li, D. Y.; Wu, Y.; Fan, R.; Yang, P. D.; Majumdar, A. Appl. Phys. Lett. 2003, 83, 3186. (7) Panev, N.; Persson, A. I.; Skold, N.; Samuelson, L. Appl. Phys. Lett. 2003, 83, 2238. (8) Poole, P. J.; Lefebvre, J.; Fraser, J. Appl. Phys. Lett. 2003, 83, 2055. (9) Bjork, M. T.; Thelander, C.; Hansen, A. E.; Jensen, L. E.; Larsson, M. W.; Wallenberg, L. R.; Samuelson, L. Nano Lett. 2004, 4, 1621. (10) Gudiksen, M. S.; Lieber, C. M. J. Am. Chem. Soc. 2000, 122, 8801. (11) Cui, Y.; Lauhon, L. J.; Gudiksen, M. S.; Wang, J. F.; Lieber, C. M. Appl. Phys. Lett. 2001, 78, 2214. (12) Ohlsson, B. J.; Bjork, M. T.; Magnusson, M. H.; Deppert, K.; Samuelson, L.; Wallenberg, L. R. Appl. Phys. Lett. 2001, 79, 3335. (13) Hiruma, K.; Yazawa, M.; Katsuyama, T.; Ogawa, K.; Haraguchi, K.; Koguchi, M.; Kakibayashi, H. J. Appl. Phys. 1995, 77, 447. (14) Martensson, T.; Borgstrom, M.; Seifert, W.; Ohlsson, B. J.; Samuelson, L. Nanotechnology 2003, 14, 1255. (15) Wu, Y. Y.; Fan, R.; Yang, P. D. Nano Lett. 2002, 2, 83. (16) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (17) Wang, J. F.; Gudiksen, M. S.; Duan, X. F.; Cui, Y.; Lieber, C. M. Science 2001, 293, 1455. (18) In a HBT measurement, when the average reciprocal count rate is much longer than the monitored time range (maximum τ), the measured n(τ) is linearly proportional to the intensity (photon) correlation function of the detected light source;1 this condition was easily satisfied in our experiments.

Nano Lett., Vol. 5, No. 7, 2005

(19) Wang, C. F.; Badolato, A.; Wilson-Rae, I.; Petroff, P. M.; Hu, E.; Urayama, J.; Imamoglu, A. Appl. Phys. Lett. 2004, 85, 3423. (20) Bayer, M.; Forchel, A. Phys. ReV. B 2002, 65, 041308. (21) Thompson, R. M.; Stevenson, R. M.; Shields, A. J.; Farrer, I.; Lobo, C. J.; Ritchie, D. A.; Leadbeater, M. L.; Pepper, M. Phys. ReV. B 2001, 64, 201302. (22) Martensson, T.; Svensson, C. P. T.; Wacaser, B. A.; Larsson, M. W.; Seifert, W.; Deppert, K.; Gustafsson, A.; Wallenberg, L. R.; Samuelson, L. Nano Lett. 2004, 4, 1987. (23) Borgstrom, M.; Deppert, K.; Samuelson, L.; Seifert, W. J. Cryst. Growth 2004, 260, 18.

Nano Lett., Vol. 5, No. 7, 2005

(24) Loss, D.; DiVincenzo, D. P. Phys. ReV. A 1998, 57, 120. (25) Piermarocchi, C.; Chen, P.; Sham, L. J.; Steel, D. G. Phys. ReV. Lett. 2002, 89, 167402. (26) Dur, W.; Briegel, H. J. Phys. ReV. Lett. 2003, 90, 067901. (27) Benjamin, S. C.; Bose, S. Phys. ReV. Lett. 2003, 90, 247901. (28) Wilson-Rae, I.; Zoller, P.; Imamoglu, A. Phys. ReV. Lett. 2004, 92, 075507.

NL050802Y

1443