NANO LETTERS
A High-Temperature Single-Photon Source from Nanowire Quantum Dots
2008 Vol. 8, No. 12 4326-4329
Adrien Tribu, Gregory Sallen, Thomas Aichele,* Re´gis Andre´, Jean-Philippe Poizat, Catherine Bougerol, Serge Tatarenko, and Kuntheak Kheng Nanophysics and Semiconductor Group, CEA/CNRS/UniVersite´ Joseph Fourier, Institut Ne´el, 25 rue des Martyrs, 38042 Grenoble cedex 9, France Received July 18, 2008; Revised Manuscript Received October 14, 2008
ABSTRACT We present a high-temperature single-photon source based on a quantum dot inside a nanowire. The nanowires were grown by molecular beam epitaxy in the vapor-liquid-solid growth mode. We utilize a two-step process that allows a thin, defect-free ZnSe nanowire to grow on top of a broader, cone-shaped nanowire. Quantum dots are formed by incorporating a narrow zone of CdSe into the nanowire. We observe intense and highly polarized photoluminescence even from a single emitter. Efficient photon antibunching is observed up to 220 K, while conserving a normalized antibunching dip of at most 36%. This is the highest reported temperature for single-photon emission from a nonblinking quantum-dot source and principally allows compact and cheap operation by using Peltier cooling.
Single semiconductor quantum dots (QDs) allow the stable generation of single photons1-3 and are compatible with current semiconductor manufacturing technology. For commercial applications, especially quantum cryptography,4 compact and easy-to-handle devices operating at room temperature are desired. For high-temperature operation, II-VI materials are especially interesting, as they can exhibit large exciton binding energies and strong carrier confinements. High-temperature experiments from individual Stransky-Krastanov (SK) grown QDs were reported from CdSe/ZnSe QDs5 and from GaN/AlN QDs.6 Both experiments showed photon antibunching up to 200 K with normalized dip values of 81% and 53%, respectively. Alternatively, a QD formed inside a semiconductor nanowire (NW) may allow design of well-positioned and sizecontrolled QDs with no wetting layer and without the necessity of self-organization mechanisms.7 Single-photon emission from a GaAsP/GaP NW QD was reported at cryogenic temperature in ref 8. Other systems have demonstrated room-temperature single-photon emission: nanocrystals9 have the drawback that they suffer from blinking effects.10 Color centers in diamond11-13 have shown reliable operation but with a broad spectrum. Anyway, neither nanocrystals nor color centers in diamond offer the possibility of electrical excitation, which is a very realistic and promising perspective for semiconducting NWs.14 Here, we report on high-temperature single-photon emission from a single CdSe QD inside a ZnSe NW. The photoluminescence (PL) * Corresponding author,
[email protected]. Present address: Institut fu¨r Physik, Humboldt Universita¨t zu Berlin, Hausvogteiplatz 5-7, 10117 Berlin, Germany. 10.1021/nl802160z CCC: $40.75 Published on Web 11/05/2008
2008 American Chemical Society
is highly polarized and shows emission from radiative excitonic decays. Photon antibunching down to 36% was observed up to a temperature of 220 K. ZnSe NWs were grown by molecular beam epitaxy (MBE) in the vapor-liquid-solid (VLS) growth mode catalyzed with gold particles. A 0.2-0.5 nm thick gold layer was deposited on a oxidized 001-Si substrate by e-beam metal deposition and dewetted by annealing the substrate at 600 °C for 5 min. Next, the samples were inserted into a II-VI MBE chamber and growth of ZnSe NWs at different sample temperatures and different Zn:Se beam-equivalent pressure ratios was performed. The growth rate is about 0.5 nm/s. Size, shape, and crystal structure are found to strongly depend on the growth conditions. Independent of the growth conditions, the NWs have a dominant wurtzite structure, while zinc blende insertions are partly observed. However, zinc blende sections are regularly observed using transmission electron microscopy (TEM).15 At low growth temperature (300 °C), or with excess of the Zn beam pressure, coneshaped NWs are formed. For higher growth temperature (>350 °C) and an excess of the Se beam pressure, the NWs have a high aspect ratio with sizes of 1-2 µm in length and 20-50 nm in width, as observed by scanning electron microscopy (SEM) and TEM.15 We have developed a two-step growth recipe to grow long and thin ZnSe NWs with a strongly reduced amount of stacking fault defects. The procedure consists of growing a narrow NW on top of the tip of a broad-based and coneshaped NW. Uniformly narrow NWs with typical diameters of 10 nm are obtained on top of broader, cone-shaped NWs,15 Figure 1. In order to prepare QDs, a small region of CdSe
Figure 1. SEM image of an as-grown NW sample. The ZnSe NWs have a broad base that tapers into a long and narrow NW including the CdSe zone. The inset shows an isolated NW on a structured substrate surface.
was inserted into this high-quality part of the ZnSe NW. The designed height of the CdSe zone is 10 nm on average, but varies between NWs as the growth rate also depends on the NW diameter. The QD size is thus of the order of the bulk exciton Bohr diameter for CdSe (11 nm). This means that the carriers in the CdSe QD are in the strong confinement regime. Additionally, two reference samples were produced both containing ZnSe only, without the inclusion of a CdSe zone: (1) with a single-step growth recipe, where a long NW is grown directly on the gold-coated substrate, and (2) with the two-step growth recipe described above. For the study of single NWs, the sample is put in a methanol ultrasonic bath for 30 s, in which some NWs broke off the substrate into the solution. This process also allows mainly the high-quality part of the narrow NW to detach from the thicker, cone-shaped part. Droplets of this solution were next placed on a clean substrate, leaving behind a low density of individual NWs. Metal markers were made on the substrate using an optical lithography technique in order to locate the NWs (inset of Figure 1). For all the optical experiments, samples were mounted on a variable-temperature cryostat allowing experiments from 4 K to room temperature (300 K). The NW emission was efficiently collected by a microscope objective and then dispersed by a monochromator. The sample was either excited by a 405 nm continuous-wave (CW) diode laser or by a 200 fs pulsed frequency-doubled Ti:Sa laser. Panels a-c of Figure 2 compare the PL of single NWs from three different samples at 4-5 K. From a single, pure-ZnSe NW grown in a single-step process, Figure 2a, we observe a broad bunch of spectral lines within 500-600 nm, emerging from excitons localized in stacking fault defects along the NW. This is in contrast to NWs grown on top a broader coneshaped NW base (Figure 2b). As these structures tend to break on the narrower NW part during the ultrasonic bath, a mostly defect-free NW remains. As seen in Figure 2b only few spectral lines remain on these NWs. The displayed ZnSe NW spectrum is one of the most intense found on the sample. Nano Lett., Vol. 8, No. 12, 2008
Figure 2. (a-c) PL spectra taken under CW excitation of a single NW structure. The sketches on the left symbolize the shape and composition of each NW (SFs ) stacking faults; // ) region where the NW is likely to break in the ultrasonic bath): (a) a pure-ZnSe NW grown in a single-step process; (b) a narrow pure-ZnSe NW grown on top of a cone-shaped NW; (c) a narrow ZnSe NW grown on top of a cone-shaped NW and including a CdSe QD zone. (d) Polarization dependence. The solid squares are the variation of the total PL intensity as a function of the polarization of the excitation laser. The open circles are the PL intensity observed through an analyzing polarizer while the excitation laser was circularly polarized. Here, the origin (0°) is arbitrarily defined as the angle of maximum intensity. (e) Radiative lifetime of an ensemble of NW structures as a function of the sample temperature. The solid blue (dashed red) curve is a linear (exponential) fit to the data points below (above) 100 K. Inset: Ensemble spectrum that we observed during the time-resolved measurement at 4 K. The green shaded area is the spectral range within which the lifetimes were evaluated.
Usually, the PL signal is much weaker. We also observe many NWs with practically no PL emission at all on this sample. With these conditions, it finally becomes possible to grow and study single NW samples with an inserted CdSe QD. The PL spectrum of such a NW is depicted in Figure 2c. Similar to single SK QDs, only a few discrete spectral lines remain. With increase in excitation power, the XX line intensity grows according to a power law with a larger exponent than the X line. This together with the measured biexciton binding energy (22 meV) is in accordance with measurements on self-assembled QDs,16 which justifies our assignment to an exciton (X) and biexciton (XX) transition. As we also find many NWs with no PL emission on this sample, the experiments below were performed on preselected NWs, in order to separate them from defective nanostructures that also appear on the substrates. Nevertheless, 1% of the NWs have well-isolated lines enabling singlephoton generation. In the following, we report solely on studies of the CdSe/ZnSe NW sample and omit the sample notation. One characteristic feature of such QD NW structures is their polarization behavior. As seen in Figure 2d, the excitation efficiency and the luminescence are both strongly 4327
polarization dependent. The PL emission is highly polarized with a contrast of 80-90% (the NW was excited here with a circularly polarized laser light). Conversely, the PL intensity has a sine-like variation as a function of the linear laser polarization. In our case the emitted light is highly polarized with the same direction as the preferred excitation polarization. In previous reports, it was found, that the polarization was highly oriented along the NW emission.17,18 This striking polarization anisotropy of absorption is explained by the dielectric contrast between the NW material and the surrounding environment; the polarization of the emitted light results from a competition between these electromagnetic effects and the orientation of the dipole within the QD.18,19 Following this argumentation, we think that in our case the polarization is aligned along the NWs, although we could not verify this alignment, due to limited optical resolution. Time-resolved measurements of the exciton lifetimes were performed using a streak camera (with a resolution of about 1 ps) on an ensemble of NWs. Figure 2e shows a flat temperature dependency of the decay time at low temperature (below 100 K) followed by an exponential decay above 100 K. The flat temperature dependency is characteristic of threedimensional confinement (confined excitons) and the exponential decay is characteristic of the nonradiative recombination regime. In the radiative regime, the decay time is about 500 ps and dominates up to 100 K. This is slightly larger than what is observed in self-assembled CdSe/ZnSe QDs (around 300 ps 20). This could be due to a piezoelectric field resulting from the wurtzite structure in the NWs, which separates electron and hole wave functions and thus reduces the oscillator strength. We have carried out photon correlation measurements using a Hanbury Brown and Twiss (HBT) setup. The total time resolution of the HBT setup was 850 ps. In order to improve the signal-to-background ratio, we made use of the strong polarization anisotropy and inserted a linear polarizer oriented for maximum transmission of the spectral line under observation. Under pulsed excitation, the second-order correlation function possesses a peaked structure indicating PL emission on demand. The graphs in Figure 2 are the raw histograms of measured coincidences without any correction for background count events. The area under each peak at τ ) 0, was normalized with respect to the average area under the peaks at |τ| > 0. Each peak area was calculated by integrating the coincidences within 12 ns windows. The correlation functions were taken at different temperatures between 4 and 220 K. At 4 K, the peak at τ ) 0 is suppressed to a normalized value of 7%, showing the high quality of the single-photon generation. With increasing temperature, this value only slightly increases to finally reach 36% at 220 K. This value is far below 50%, the emitted light field is thus clearly distinguished from states with two or more photons. Thus, even without correcting for background events, these emitters can be directly used as a high-quality single-photon device with a strongly suppressed probability for two-photon events, even when operating at high temperature. This was not reported before from a nonblinking semiconductor system. 4328
Figure 3. (a) Second-order correlation taken under pulsed excitation at temperatures between 4 and 220 K. The numbers in the graphs are measured values of g(2)(0) (i.e., the area under the peak at τ ) 0 relative to the peaks at |τ| > 0). The numbers in parentheses are the reconstructed values of the pure spectral lines, gS(2)(0). (b) Spectra from the same NW QD taken during the same experimental run as Figure 3a.
Figure 3b shows the spectra from the NW QD used for the correlation measurements in Figure 3a. With increase in temperature, the spectrum evolves from a dominant trion and a weak exciton transition at 4 K to a broad biexciton line above 150 K. The increased biexciton intensity at high temperature is due to an higher excitation power, which we adjusted at each temperature to achieve an optimized output of the correlation function. Moreover, at high temperature, the biexciton is less affected by nonradiative decay than the exciton due to the shorter biexciton lifetime and due to coupling of bright and dark excitons over spin flips. The assignment of the lines to the decay of an exciton, trion, and biexciton was obtained from power dependencies of the line intensities and from intensity cross-correlations (not shown here). A detailed study of the spectral behavior as a function of the temperature is in process and is beyond the scope of this paper. The analysis of the spectra also allows the contribution of the observed spectral line S to separate from its spectral background B. These values can be assessed from integrating the areas under the peak and the spectral background over the corresponding spectral window of the spectrometer output slit. The correlation function of the pure spectral line gS(2)(τ) can be reconstructed from the measured correlation function via gS(2)(τ) ) 1 + (g(2)(τ) - 1)/F2, with F ) S/(S + B).11 The values for gS(2)(0) are given in Figure 3a in parentheses and range within 1%-11%. This shows that the observed spectral lines indeed originate from a single transition. Although the quality of the single-photon statistics remains mostly constant, the count rate decreases from 25000 counts/s at 4 K to 5000 counts/s at 220 K and the width of the spectral lines significantly broadens. In the correlation measurements obtained at 190 and 220 K, a low-resolution grating was used in the spectrometer so that all the photons coming from the broader line could be counted. This broader spectral window Nano Lett., Vol. 8, No. 12, 2008
of integration leads to larger background, which is the main origin for the rise of the τ ) 0 peak above 150 K. In contrast to SK QDs, which often grow with a high density on the substrate, the density of NWs in the microscope focus was much smaller and can be even reduced to only one within the microscope focus, which avoids contributions from neighboring emitters that spectrally overlap with the transition under observation. However contributions from other transitions of the same NW cannot be excluded. In summary, we have performed optical studies on single CdSe QDs inside ZnSe NWs. The NWs were developed by MBE in a two-step growth recipe, where narrow, mostly defect-free NWs are grown on top of broader, cone-shaped NWs. The single-NW PL is highly polarized with a contrast of 80-90% and features spectral lines from exciton, biexciton, and trion transitions with lifetimes of around 500 ps. When individual transitions were filtered, nonclassical singlephoton statistics were retrieved, indicated by strong antibunching, where the raw correlation function g(2)(τ ) 0) was reduced down to a normalized value of 7%. This behavior remains even up to a temperature of 220 K, where this correlation peak is only slightly increased to 36%. For nonblinking QDs, this is the highest reported temperature for single-photon emission and for an antibunching dip below 50%. At this temperature, Peltier cooling becomes an alternative to liquid helium or nitrogen cooling. Together with the possibility of integrating NWs into electro-optical circuits,14 these emitters become an interesting candidate for developing compact, stable, and cost-efficient quantum devices operating near room temperature. Acknowledgment. T.A. acknowledges support by Deutscher Akademischer Austauschdienst (DAAD). Part of this work was supported by European project QAP (Contract No. 15848).
Nano Lett., Vol. 8, No. 12, 2008
References (1) Michler, P. Kiraz, A. Becher, C. Schoenfeld, W. V. Petroff, P. M. Zhang Lidong, Hu E. Imamoglu, A. Science 2000, 290, 2282. (2) Santori, C.; Pelton, M.; Solomon, G.; Dale, Y.; Yamamoto, Y. Phys. ReV. Lett. 2001, 86, 1502. (3) Articles in Focus Issue: Single Photons on Demand. New J. Phys. 2004, 6. (4) Gisin, N.; Ribordy, G.; Tittel, W.; Zbinden, H. ReV. Mod. Phys. 2002, 74, 145. and citations therein. (5) Sebald, K.; Michler, P.; Passow, T.; Hommel, D.; Bacher, G.; Forchel, A. Appl. Phys. Lett. 2002, 81, 2920. (6) Kako, S.; Santori, C.; Hoshino, K.; Go¨tzinger, S.; Yamamoto, Y.; Arakawa, Y. Nat. Mater. 2006, 5, 887. (7) Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, F.; Yan, H. AdV. Mater. 2003, 15, 353. (8) Borgstro¨m, M. T.; Zwiller, V.; Mu¨ller, E.; Imamoglu, A. Nano Lett. 2005, 5, 1439. (9) Michler, P.; Imamoglu, A.; Mason, M. D.; Carson, P. J.; Strouse, G. F.; Buratto, S. K. Nature 2000, 406, 968. (10) Brokmann, X.; Hermier, J.-P.; Messin, G.; Desbiolles, P.; Bouchaud, J.-P.; Dahan, M. Phys. ReV. Lett. 2003, 90, 120601. (11) Brouri, R.; Beveratos, A.; Poizat, J.-P.; Grangier, P. Opt. Lett. (2000) , 25, 1294. (12) Kurtsiefer, C.; Mayer, S.; Zarda, P.; Weinfurter, H. Phys. ReV. Lett. 2000, 85, 290. (13) Wu, E.; Rabeau, J. R.; Roger, G.; Treussart, F.; Zeng, H.; Grangier, P.; Prawer, S.; Roch, J.-F. New J. Phys. 2007, 9, 434. (14) Minot, E. D.; Kelkensberg, F.; van Kouwen, M.; van Dam, J. A.; Kouwenhoven, L. P.; Zwiller, V.; Borgstro¨m, M. T.; Wunnicke, O.; Verheijen, M. A.; Bakkers, E. P. A. M. Nano Lett. 2007, 7, 367. (15) Aichele, T.; Tribu, A.; Bougerol, C.; Kheng, K.; Andre´, R.; Tatarenko, S. Appl. Phys. Lett. 2008, 93, 143106. (16) (a) Tu¨rck, V.; Rodt, S.; Heitz, R.; Stier, O.; Strassburg, M.; Pohl, U. W.; Bimberg, D. Phys. Status Solidi b 2001, 224, 217. (b) Patton, B.; Langbein, W.; Woggon, U. Phys. ReV. B 2003, 68, 125316. (17) Lan, A.; Giblin, J.; Protasenko, V.; Kuno, M. Appl. Phys. Lett. 2008, 92, 183110. (18) Wang, J.; Gudiksen, M. S.; Duan, X.; Cui, Y.; Lieber, C. M. Science 2001, 293, 1455. (19) Niquet, Y. M.; Mojica, D. C. Phys. ReV. B 2008, 77, 115316. (20) Bacher, G.; Weigand, R.; Seufert, J.; Kulakovskii, V. D.; Gippius, N. A.; Forchel, A. Phys. ReV. Lett. 1999, 83, 4417.
NL802160Z
4329
