Photoluminescence of GaN Nanowires of Different Crystallographic Orientations

We utilized time-integrated and time-resolved photoluminescence of a-axis and c-axis gallium nitride nanowires to elucidate the origin of the blue-shifted ultraviolet photoluminescence in a-axis GaN nanowires relative to c-axis GaN nanowires. We attribute this blue-shifted ultraviolet photoluminescence to emission from surface trap states as opposed to previously proposed causes such as strain effects or built-in polarization. These results demonstrate the importance of accounting for surface effects when considering ultraviolet optoelectronic devices based on GaN nanowires. GaN nanowires have been the subject of intense research lately due to the many potential applications (lasers, lightemitting diodes, modulators, detectors, etc.)1-3 and interesting properties4,5 that they possess. Because GaN has a wurtzite crystal structure, which is an anisotropic crystal structure, many of its properties are dependent upon crystal orientation. For example, the photoluminescence (PL) of GaN nanowires in the ultraviolet (UV) has been observed to depend on the growth direction of GaN.4,5 However, the origin of the difference in PL between nanowire samples of different growth directions remains unclear. Understanding the cause of this difference can have significant impact on improving the performance of future optoelectronic devices based on GaN nanowires, as the nature of the different PL sources (e.g., band edge, defects, etc.) can impact device performance (e.g., gain).6 Previous attempts to explain the dependence of PL on GaN growth direction have invoked various phenomena: strain in the nanowires,7 dopants,8 or built-in polarization.5 The spontaneous built-in polarization, which is responsible for the PL peak shift in quantum wells,9 can be ruled out. In quantum well growth, the growth perpendicular to the (001) polar surface (c-plane) will result in a built-in field, which leads to red-shift of the PL peak. In the case of nanowires, because the field along wire direction can be ignored due to the relatively large distance (∼10 μm), growth along the 〈001〉 direction (along the c-axis) will not result in a significant built-in field. Rather, growth along the 〈1-10〉 direction (along the a-axis) could result in a large built-in potential, but the blue-shift of PL in the a-axis nanowire is not consistent with such an explanation. Given that the surface plays such an important role in nanoparticles (due to the high surface-to-volume ratio) and the majority of the luminescent material is in proximity to the surface (half of the volume of a 20 nm diameter nanowire is within ∼3 nm of the surface), perhaps the nature of the surfaces (termination, shape, strain, defects, etc.) can account for the difference, as has been suggested by other studies.10-13 In the extreme case of spherical nanoparticles, it is now wellestablished that oxide surface-state luminescence can significantly contribute to the PL.12,13 To determine if surface states play a role in the dependence of GaN nanowire photoluminescence on crystal orientation, we use timeintegrated photoluminescence (TIPL) and time-resolved photoluminescence (TRPL) to study the PL from GaN nanowire samples of different crystallographic orientation. We utilized two different UV PL setups to study the TIPL of GaN nanowires. One setup was a commercial Raman * Corresponding authors. E-mail: [email protected] (A.H.C.); [email protected] (C.-Z.N.). † Center for Advanced Aerospace Materials and Devices, NASA Ames Research Center. ‡ Department of Chemistry, University of California. § Department of Chemical Engineering, University of Louisville. | Current address: Department of Materials Science and Engineering, Lehigh University, Bethlehem, Pennsylvania 18015. ⊥ Current address: Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139. NANO LETTERS 2007 Vol. 7, No. 3 626-631 10.1021/nl062524o CCC: $37.00 © 2007 American Chemical Society Published on Web 02/27/2007 D ow nl oa de d by U N IV O F L O U IS V IL L E o n Ju ly 1 , 2 00 9 Pu bl is he d on F eb ru ar y 27 , 2 00 7 on h ttp :// pu bs .a cs .o rg | do i: 10 .1 02 1/ nl 06 25 24 o scattering and PL system (Renishaw inVia Raman microscope) that uses a 325 nm continuous-wave (CW) He-Cd UV laser as an excitation source. The PL was collected by an objective lens, directed to a spectrometer, and detected by a thermoelectrically cooled CCD detector. While a laserheating-induced red-shift was observed at the highest excitation intensities with this setup, the data shown here are with low excitation intensity to avoid laser-heating effects. The other setup was based on a passively mode-locked Ti: sapphire laser (810 nm, ∼150 fs pulse duration, 80 MHz repetition rate, 1.5 W average power) that was frequency tripled to yield ∼100 mW average power of light at 271 nm. This pulsed (∼200 fs pulse duration) UV beam was focused by a 500 mm focal length lens and directed to the sample at ∼70° from normal, yielding a spot size of ∼60 μm × 180 μm. This led to a maximum aVerage intensity of ∼900 W/cm2 and a maximum peak intensity of ∼6 × 107 W/cm2 per pulse on the sample. The PL was collected by a UV objective (13×), directed into a 0.3 m spectrometer, and detected with a liquid-nitrogen-cooled CCD detector. The spectral resolution of this system can be as high as ∼0.2 nm. For TRPL, we used ultrashort pulses at 267 nm obtained from the frequency-tripled output of a 40 kHz regeneratively amplified Ti:sapphire laser system with a 150 fs pulse width. The maximum aVerage laser intensity on the sample is less than 400 W/cm2. The PL of the samples was collected at 45° relative to the excitation beam. The PL is collimated and focused into a spectrometer to disperse the spectrum before being passed to a picosecond streak camera (Hamamatsu C4334 Streakscope). The instrument response is ∼15 ps in a 1 ns sweep window, and the spectral resolution is 2