Stimulated emission from ZnO nanorods

Transcription

1 phys. stat. sol. (b) 243, No. 4, (2006) / DOI /pssb Editor s Stimulated emission from ZnO nanorods R. Hauschild *,1, H. Lange 1, H. Priller 1, C. Klingshirn 1, R. Kling 2, A. Waag 3, H. J. Fan 4, M. Zacharias 4, and H. Kalt 1 1 Institut für Angewandte Physik, Universität Karlsruhe (TH), Karlsruhe, Germany 2 Abteilung Halbleiterphysik, Universität Ulm, Albert-Einstein Allee 45, Ulm, Germany 3 Institut für Halbleitertechnik, TU-Braunschweig, H.-Sommer-Str. 66, Braunschweig, Germany 4 Max-Planck-Institut für Mikrostrukturphysik, Weinberg 2, Halle, Germany Received 12 September 2005, revised 15 October 2005, accepted 3 January 2006 Published online 2 February 2006 PACS h, p, Et, n By means of time resolved spectroscopy we compare two samples of ZnO nanorods with respect to their suitability as stimulated emitters. In the case of narrow nanorods their wave guiding quality causes a suppression of exciton exciton scattering whereas no laser emission is detectable. Unlike their narrow counterparts, wide nanorods not only benefit from a larger overlap of the guided mode with the gain medium but a variation in VLS growth results in gold nanoparticles being present at the bottom of nanorods. Consequently, laser emission from single wide rods is evidenced up to 150 K. In addition to experimental studies we carry out 3D numerical simulations of the electric field distribution to evaluate the influence of gold nanoparticles at the nanorod/substrate interface. This finite element analysis confirms that gold leads to an enhancement of confinement within the resonator. 1 Introduction The large band gap semiconductor zinc oxide (ZnO) is of significant interest for optoelectronic application in the ultraviolet. Recent progress in temperature modulation epitaxy [1] opens a possible road towards reliable p-doping of ZnO which in turn carries the promise of electrically pumped ZnO nanorod lasers. We report on investigations of stimulated emission of single nanorods and on numerical simulations concerning the influence of geometric parameters on the suitability of nanorods as laser emitters. 2 Experimental details 2.1 Growth of ZnO nanorods ZnO nanorods of high structural quality can be prepared in various ways. We compare rods grown by a modified vapor liquid solid (VLS) process [2] to rods grown by metal-organic vapor phase epitaxy (MOVPE) [3]. These samples differ mainly in their rod diameter which has a pronounced effect on their optical properties. The fabrication process of the VLS samples is template assisted and involves gold nanodisk arrays obtained by using a gold membrane. The membrane itself is a replica of an amorphous aluminum oxide membrane as a deposition mask. The gold nanodisk array allows a subsequent sitespecific growth of ZnO on a GaN-covered Si substrate. The result is a long range ordered array of ZnO * Corresponding author: robert.hauschild@physik.uni-karlsruhe.de, Phone: , Fax:

2 854 R. Hauschild et al.: Stimulated emission from ZnO nanorods Fig. 1 Scanning electron microscopy images of a MOVPE-grown (left) and a template-assisted VLSgrown sample (right). nanorods with a narrow size distribution centered at a diameter of 300 nm and a typical height of 1.5 µm, i.e., an aspect ratio of 5 (wide rods). The high growth temperature of 800 C at the substrate leads to a deviation from the conventional growth mechanism resulting in a thin interface layer of dispersed gold particles at the bottom of the nanorods [4]. The MOVPE samples are grown without any gold catalyst directly on sapphire a-plane substrates. The rods investigated here have a diameter of 50 nm and a length of 5 µm, i.e., an aspect ration of 100 (narrow rods). Rods grown with both techniques have a hexagonal cross section in growth direction and the crystal faces are aligned. Scanning electron microscopy pictures of both sample types are shown in Fig Set-up of optical experiments Photoluminescence (PL) spectra are obtained using a time-resolved micro-pl system (temporal resolution: 5 ps, spectral resolution: 0.5 mev, spatial resolution: <500 nm). A frequency doubled Ti:Sapphire laser delivers ~100 fs pulses at 3.49 ev with an average power of up to 100 mw. The PL signal is spectrally dispersed and recorded with a combination of synchroscan streak camera and CCD array. The sample is mounted in a He flow cryostat for temperature dependent measurements. The UV microscope objective (magnification: 20, numerical aperture 0.4) is placed outside the cryostat. This set-up allows global excitation of an ensemble of nanorods as well as a tight focusing of the excitation spot in a confocal arrangement. The spatial resolution in the later case allows addressing a group of few narrow nanorods or single wide nanorods, respectively. 3 Stimulated emission from single nanorods Both, narrow and wide nanorods show bulk-like excitonic photoluminescence at low excitation reflecting the high crystalline quality. The PL decay time increases with temperature as expected for 3D excitons [5]. In the intermediate excitation regime one observes non-linear contributions to the PL signal that differ between the two samples. The P-band emission resulting from exciton exciton scattering which is very pronounced in ZnO epitaxal layers and bulk samples is weaker in wide nanorods (see Fig. 2) and negligible in narrow nanorods. The possibility that this absence of the P-band is a peculiarity of this specific sample can be ruled out since this effect has also been seen in a variety of other narrow nanorod samples. We attribute this observation to the reduced phase-space for polariton polariton scattering since the light wave coupling to the 3D exciton has a 1D confinement in the rods (see next chapter). Increasing the excitation fluence in the confocal arrangement leads to the onset of stimulated emission in single wide rods (Fig. 2). Such a transition does not occur in the narrow nanorods even under global excitation. The spectral position of the stimulated recombination varies from rod to rod. Together with

3 phys. stat. sol. (b) 243, No. 4 (2006) 855 PL Intensity (arb. u.) E-3 Excitation Fluence: 850µJ/cm 2 160µJ/cm 2 20µJ/cm 2 T=24K PL Intensity (arb. u.) E-4 X-LO P P Photon Energy (ev) X A Excitation Fluence (µj/cm 2 ) Fig. 2 PL spectra (left) from a single wide nanorod showing P-band emission at intermediate excitation level and the onset of stimulated emission. The PL intensity displays a threshold-like behavior as a function of excitation fluence (right). the quenching of the P-band emission we can therefore exclude a pure excitonic origin. Instead free carriers have to be involved. The stimulated emission is present under femtosecond, confocal pumping up to 150 K. We can exclude that the observed differences in the occurrence of stimulated emission is related to sample quality. Rather, the behavior at intermediate excitation levels infers that the waveguide and resonator properties of the nanorods have a significant influence on the PL. We will discuss this influence in the following. 4 Resonator and waveguide properties of nanorods The resonator and waveguide properties are investigated with two different approaches. A first estimate of the resonator Q-factor is given by a simple Fabry Perot model of the resonator. In this model we assume an index of refraction of ZnO of 2.2 (n GaN = 2.4) which leads to a Q-factor close to unity. Even under the assumption of a refractive index of n = 9.8 [6] for ZnO (this would be the value close to the excitonic resonance at low temperatures) we find a Q-factor of the wide-rod resonators of at most 20. Under realistic conditions (high density of free carriers due to optical pumping) one finds the excitonic resonance strongly dampened and the effective refractive index significantly reduced. In the case of micro-resonators losses not only result from the reflectivity of the end mirrors but also from their limited size. Hence, the Fabry Perot model gives an upper estimate of the Q-factor and we can conclude that ZnO rods on the bare substrate alone are bad resonators. More detailed and accurate evaluation is possible from a numerical evaluation of the scalar [7] Helmholtz equation using the finite element method for the discretization (3D modeling). In Fig. 3a) we show in the left panel the waveguide modes for a ZnO rod on GaN substrate. It is obvious, that a significant leakage of the resonator modes into the substrate occurs due to the small step in the refractive index. An evaluation of the modes as a function of rod size shows that an increasing fraction of the mode extends laterally into the vacuum when the rod diameter is reduced. This behavior can explain the missing stimulated emission from the narrow rods since the overlap of the resonator mode with the gain medium is reduced. The fact that we have stimulated emission from the wide rods is then a result of the better mode guiding but is also influenced by the presence of gold at the resonator facets. The left panel in Fig. 3 shows the influence of a thin gold layer on the resonator properties. Leakage into the substrate is significantly reduced. The index of refraction of gold required for the FEM calculation is derived from Drude s model

4 856 R. Hauschild et al.: Stimulated emission from ZnO nanorods Fig. 3 a) Intensity distribution of the electric field inside the modeled volume. As shown in the right part of a), the higher step in index of refraction at the interface gold/gan leads to an improved confinement of light inside the nanorod and to less leakage into the substrate. b) Cross section perpendicular to the growth direction. c) TEM picture of the actual interface nanorod/substrate. The gold particles appear as dark droplets [4]. assuming that it still holds with a scattering frequency that is twice that of bulk gold due to the additional scattering at the surfaces [8]. We conclude that wide nanorods offer better overlap of the resonator eigenmodes with the gain medium. Leakage of these resonator modes into the substrate can be significantly reduced by the presence of gold which is achieved readily by a high substrate temperature during growth. This finding leads the way to an improvement of ZnO nanorod resonators. Acknowledgement We acknowledge financial support by the Landesstiftung Baden-Württemberg within the Competence Network Functional Nanostructures, Project A1. References [1] A. Tsukazaki, A. Ohtomo, T. Onuma, M. Ohtani, T. Makino, M. Sumiya, K. Ohtani, S. Chichibu, S. Fuke, Y. Segawa, H. Ohno, H. Koinuma, and M. Kawasaki, Nature Mater. 4, 42 (2005). [2] H. J. Fan, F. Bertram, A. Dadgar, J. Christen, A. Krost, and M. Zacharias, Nanotechnology 15, 1401 (2004). [3] R. Kling, C. Kirchner, T. Gruber, F. Reuss, and A. Waag, Nanotechnology 15, 1043 (2004). [4] H. J. Fan, W. Lee, R. Hauschild, M. Alexe, R. Scholz, A. Dadgar, K. Nielsch, H. Kalt, A. Krost, M. Zacharias, and U. Gösele, Small, submitted (2005). [5] H. Priller, R. Hauschild, J. Zeller, C. Klingshirn, H. Kalt, R. Kling, F. Reuss, Ch. Kirchner, and A. Waag, J. Lumin. 112, 173 (2005). [6] Landolt-Börnstein: New Series III, Vol. 34C2 (Springer, Berlin, 2004). [7] H. Li, Vertical cavity surface emitting laser devices (Springer, New York, 2003), p [8] S. Linden, C. Enkrich, M. Wegener, J. Zhou, T. Koschny, and C. Soukoulis, Science 306, 1351 (2004).

5 phys. stat. sol. (b) 243, No. 4 (2006) 857 Robert Hauschild studied physics at the Technical University Munich, the University of Oregon and the University of Konstanz where he earned his degree. After that he received a scholarship of the Graduate College Collective Phenomena in Solids to pursue a Ph.D. at the University of Karlsruhe (TH) where he is currently employed. Among his research interests are the optical properties of nano- and mesoscopic systems. One part of his ongoing investigations is concerned with excitonic processes in ZnO nanorods and waveguides. The second part of his work deals with energy transfer in artificial mimics of natural light harvesting complexes. He is looking forward to extending his knowledge into the field of biology and to acquiring new skills.