Two-Photon Photoluminescence from Hierarchical ZnO Nanostructures

ECS Transactions, 45 (5) 67-72 (2012) 10.1149/1.3700411 © The Electrochemical Society

Two-Photon Photoluminescence from Hierarchical ZnO Nanostructures G. Grinblata,b,d, M.G. Capelutoa, M. Tiradoc, D. Comedib,d, A.V. Bragasa,e a

Laboratorio de Electrónica Cuántica, Depto. de Física, FCEyN, UBA, CABA, Argentina. b Laboratorio de Física del Sólido, Depto. de Física, FACET, UNT, S.M. de Tucumán, Argentina. c Laboratorio de Propiedades Dieléctricas de la Materia, Depto. de Física, FACET, UNT, S.M. de Tucumán, Argentina. d CONICET, Argentina. e IFIBA-CONICET-UBA, Argentina. Hierarchical ZnO nanostructures (such as nanowires, nanocombs and nanohedgehogs) were grown by means of a vapor-phase transport technique in a tubular furnace. Silicon based substrates were used, over which Au nanoclusters were deposited as growth catalyzers. We study the two-photon absorption induced photoluminescence (TPPL) from the various ZnO nanoobjects grown using a micrometer sized polarised laser beam as the excitation source. The UV region of the TPPL spectra contains various peaks whose different patterns depend on the type of nanostructure and its orientation with respect to the incident polarization. We identify the peaks in the TPPL spectra, which are found to be dominated by free exciton emission and their phonon replicas. For the nanocomb, however, Fabry-Perot modes originating from optical multiple reflections within the comb’s teeth are demonstrated.

Introduction ZnO is a wide (∼3.3 eV at room temperature), direct band gap semiconductor with a high exciton binding energy (∼60 meV), hence a promising material for applications in optoelectronic devices, such as ultraviolet laser diodes, solar cells, ultrasensitive gas sensors and many others (1). In this work, we present two-photon absorption induced photoluminescence (TPPL) and non-resonant Raman scattering measurements on a variety of ZnO nanostructures obtained by a Au-catalized vapour-transport deposition (VTD) technique.

Experimental Details The ZnO nanostructures were fabricated by VTD within a quartz tube under a ultrahigh purity Ar+O2 flux using a tubular furnace system and method described elsewhere (2). Substrates were Au nanocluster-covered quartz and thermally oxidized and bare Si wafers. The Au nanoclusters on the substrates are needed to promote the nanostructure growth and were obtained by depositing a nominally 1 nm thick Au film over the substrates. The Ar and O2 flow rates were separately controlled to obtain desired O2/Ar

67 Downloaded 27 Apr 2012 to 93.206.60.180. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

ECS Transactions, 45 (5) 67-72 (2012)

flow-rate ratios using independent mass-flow controllers. Further details on the growth of the ZnO nanostructures studied here will be presented in a forthcoming publication. The scheme of the TPPL experimental setup is shown in Figure 1. A Ti:Sapphire (Ti:Sa) pulsed laser tightly focuses onto the sample surface through a long working distance microscope objective (20X Mitutoyo Plan Apo Infinity), after the beam expansion with a telescope (L1-L2). A mirror M is used to deflect the beam toward the sample, so that it impinges at gracing incidence (60°-70° with respect to the substrate normal). The scattered light is collected in back-scattering geometry by two lenses (L3 and L4) and sent to a double grating Raman spectrometer (Jobin Yvon U1000) with liquid nitrogen cooled CCD camera detection. Ti:Sapphire (Ti:Sa) oscillator produces linearly polarized pulses of 70 fs pulse width and 90 MHz repetition rate which can be tuned up to a central wavelength of 740 nm. Raman measurements have been taken using the 514.5 nm line of an Ar+ ion laser. Raman spectroscopy of all the nanostructures resulted in an intense peak at 440 cm-1, assigned to the E2 longitudinal optical phonon (LO). Morphology and density of the nanostructures were studied with a scanning electron microscope (field emission SEM, Supra 40, Zeiss).

Figure 1. Schematic of the experimental set up: M mirror, L lenses. A telescope composed by L1 and L2 expands the beam at the entrance of a microscope objective, which focuses it onto the sample surface through the mirror M. The scattered light is collected in back scattering by two lenses (L3 and L4).

Results and Discussion Figure 2(a) shows a SEM image, taken normal to the sample surface, of ∼50 nm diameter, ∼5 µm long nanowires, grown from Au nanoclusters sputtered onto a quartz substrate (nominal diameter size: 5 nm). We estimate to have around 10 nanowires into the laser beam spot (∼10 µm2) on this sample. The nanowires are preferentially aligned in the direction perpendicular to the substrate, which we infer from a SEM image taken parallel to the sample surface (not shown here). Figure 2(b) shows the TPPL measurement in this sample. An intense green band assigned to the luminescence coming from surface defects (3) is observed together with a less intense structured peak in the UV region, associated with exciton emission from the ZnO. The inset shows in more detail the UV TPPL spectrum and its dependence on the incident beam polarization. When the light polarization has a significant component parallel to the nanowires (caxis), three peaks can be observed at 370 nm (3.351 eV), 375 nm (3.307 eV) and 382 nm (3.246 eV) together with a broader emission centered about 390 nm (3.18 eV). When the polarization is turned perpendicular to the c-axis, only the second peak remains along with the broad emission. The peak at 375 nm is the free exciton (FX) emission, which is collected in both polarizations, suggesting that it is not a pure guided mode on the

68 Downloaded 27 Apr 2012 to 93.206.60.180. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

ECS Transactions, 45 (5) 67-72 (2012)

nanowire. The peak at 370 nm is the second harmonic signal (SH), and the peak at 382 nm is assigned to the phonon replica (FX-1LO); both are only efficiently generated for the parallel polarization. The frequency shift between FX and FX-1LO, coincides with the LO phonon measured by non-resonant Raman spectroscopy. The SEM images showed a ∼1 μm thick ZnO wetting layer over which the nanowires grew. Then, the broad peak centered at 390 nm, which does not change with the incident polarization, may possibly be regarded as the emission from this ZnO bottom layer. Since this sample has a low number of nanowires per unit area (∼1 nanowire/μm2), a relevant amount of the excitation beam reaches the bottom layer producing significant luminescence from that region. Emission of higher order phonon replicas, spread by the slightly different directions of the nanowires into the laser spot, could also contribute to this broad peak. ZnO has a very efficient LO phonon-exciton coupling and, therefore, the excitonpolariton formed after the two-photon absorption may relax emitting an LO phonon. The resulted photon-like polariton have a large and well-defined c-component wavevector and can be confined to propagate along the nanowire axis with very small damping (4). 400

50

(a)

(b)

Incident polarization

320

IPL (a.u.)

40

IPL (a.u.)

240

30 20 10 0 360

370

380

390

400

410

420

λ (nm)

160

80

0 360

400

440

480

520

560

600

640

λ (nm)

Figure 2. (a) SEM image taken perpendicular to the sample surface shows the ZnO nanowires grown. (b) TPPL spectrum from ZnO nanowires. The inset shows the dependence on the incident light polarization. S-polarized incoming beam with a component parallel to the nanowires (upper curve) shows: SH (370 nm), FX mode (375 nm), FX-1LO mode (382 nm). P-polarization perpendicular to the nanowires (lower curve) shows only the FX mode (375 nm). These results suggest that FX-1LO is a guided mode in the nanowire cavity (see text). Figure 3 shows the TPPL for nanohedgehogs deposited directly on a Si(100) substrate (without gold seeds). A nanohedgehog is composed by about fifteen nanowires of 150 nm in width and up to 10 µm in length which grow from a ∼4 µm3 central core. Since the laser spot has an area of 10 µm2, then, on average, 1 nanohedgehog enters into the spot, and consequently the TPPL signal corresponds to a one single nanostructure. A strong FX emission at 375 nm (3.307 eV) is observed, whereas the SH and FX-1LO contributions are not detected for this nanoobject. This observation is compatible with the fact that the nanowires are evenly distributed around the core and a preferred direction of growth does not exist. Also, the defect-related visible emission is totally absent in this sample, since the surface to volume ratio for this sample is relatively small.

69 Downloaded 27 Apr 2012 to 93.206.60.180. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

ECS Transactions, 45 (5) 67-72 (2012)

120

(a)

(b) 100

IPL (a.u.)

80 60 40 20 0 360

380

400

420

440

460

480

500

λ (nm)

Figure 3. (a) SEM image of the sample. The inset shows a closer view from which the shape of the nanohedgehogs can be appreciated. (b) TPPL spectrum from a ZnO nanohedgehog, free exciton emission (FX) at 375 nm is observed without any emission in the visible. A silicon wafer covered by 500 nm of amorphous silicon oxide was used as a substrate to grow nanocombs, nanowires and nanosheets using nanometric size gold islands as seeds. Figure 4 shows the SEM image of a sample with nanocombs (a) and the TPPL spectrum for a single structure (b). The mean length of one comb is greater than 5 µm and each tooth has a length of around 1 µm. Therefore, 1 or 2 of these combs occupy the area of the laser spot, and consequently we can assume that we are taking the spectrum of a single structure. Highly structured TPPL spectra in the UV region and a broad defect-related visible emission band are observed. The UV spectrum is composed of both structured TPPL and SH radiation. 90

(a)

(b) 75

IPL (a.u.)

60 45 30 15 0 360

400

440

480

520

560

600

640

λ (nm)

Figure 4. (a) SEM image and (b) PL spectrum from ZnO nanocombs. Highly structured TPPL is observed due to Fabry-Perot optical modes and phonon replicas. Figure 5(a) shows the UV region of the spectrum at different locations of the sample. As it can be seen from the figure, different peaks appear at different sample locations. As an example, FX mode and FX-1LO phonon replica are both easily observed in the upper curve but FX and FX-1LO are not present in the middle and lower curves respectively. The phonon replicas will be more visible when the comb teeth are aligned with the

70 Downloaded 27 Apr 2012 to 93.206.60.180. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

ECS Transactions, 45 (5) 67-72 (2012)

incident polarization. As the combs are randomly oriented over the sample, this will happen only for certain locations over the sample. Besides the FX modes and phonon replicas, we observe modes with smaller spacing that corresponds to Fabry-Perot optical modes. This is demonstrated in Figure 5(b), which shows the frequency of the maxima as a function of the mode order, for the three spectra in Figure 5(a) (taking an arbitrary origin for the order number). We observe a linear behavior in accordance with the frequency modes of a Fabry-Perot cavity. The mode spacing in a Fabry-Perot interferometer is Δν = c2 / 2d where c2 ≈ 107 m / s (5) is the polariton propagation velocity in the c-axis for ZnO and d is the size of the resonator, being in our case the length of the comb’s tooth. We relate the slope of the linear fittings with the frequency separation in order to obtain the length of the resonator, which results to be d ∼1.5 to 1.6 µm. This result is in accordance with the lengths measured at the SEM image (Figure 4(a)). 200

(a)

SH

795 FX

FX-LO

FX-2LO

(b) frequency (THz)

160

IPL (a.u.)

I 120 80

II

40

790

phonon replicas

785 780

slopes 3.34 THz (I) 3.04 THz (III) 3.34 THz (II)

775

I III II

770

III

0 365

370

375

380

385

390

765

395

λ (nm)

1

2

3

4

5

6

7

8

order number

Figure 5. (a) TPPL and SH at different locations (I, II, III) of the sample. (b) Frequency at the maxima position as a function of the mode order for the curves shown in (a). The linear dependence with the mode order indicates the presence of Fabry-Perot modes in the comb’s tooth. The numbers in the label of (b) indicate the slopes of the linear fits. In order to establish stationary modes in a Fabry-Perot resonator, both ends of the cavity should be parallel and high optical quality reflectors. Scattering at the ends of the cavity reduces the reflectivities and, therefore, also the multiple reflections of the photon field from internal surfaces of the nanostructures that are needed to build up a stationary wave. For a single tooth of a comb, both ends of the cavity have enough optical quality surfaces to act as high reflectors. One end, equivalently to the nanowires and nanohedgehogs, is crystalline hexagonal shaped (observed by SEM), whereas the other one is attached to a bigger crystalline structure (the body of the comb). Then, at this junction, the optical waves see a change of impedance due to the large area difference between the comb’s tooth and body cross sections, leading to a great reflectivity. It should be stressed that, for the Fabry-Perot modes be detectable, the nanostructure system needs to be as monodisperse as possible (i.e. all resonating cavities should have similar lengths and orientations with respect to the beam polarization). For a higly polydisperse system, such as the randomly oriented nanowires in Ref. (2), the fringes would be blurred as a result of a wide phase difference distribution. This may be the reason why we do not see Fabry-Perot modes in the other nano-objects studied here (i.e., nanowires and nanohedgehogs) where the laser spot probes various nanoobjects having different lengths

71 Downloaded 27 Apr 2012 to 93.206.60.180. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

ECS Transactions, 45 (5) 67-72 (2012)

and/or directions, rendering the effect unobservable. In addition, the surface of the wires attached to the substrate or to the nanohedgehog core may not have enough optical quality surfaces to produce Fabry-Perot modes, in contrast to the case of the comb’s teeth (4).

Conclusions In this work we identified the different peaks present in the UV part of the TPPL spectra for a variety of ZnO nanostructes fabricated through a vapour-transport deposition technique. We found emission associated with the free exciton (FX) and the phonon replicas (FX-nLO). The phonon frequency was first determined through Raman measurements. For quasi-oriented nanowires, when the light polarization has a significant component parallel to them, second harmonic generation is produced and first order phonon replica appears as a guided mode in the nanowire cavity. These two contributions of the spectra disappear for a perpendicular polarization. This is why for nanohedgehogs, which have nanowires equally distributed radially to the cores, neither second harmonic generation nor phonon replicas were observed. In the case of nanocombs, we found the teeth to be good optical resonators that support Fabry-Perot modes. We calculated the lengths of the cavities from TPPL measurements on different nanocombs and found them to agree with the mean length of the teeth measured by SEM. Since the nanocombs are randomly distributed in orientation over the sample, only in some regions of it phonon replicas could be measured, that is, when the nanocomb probed by the laser spot had its teeth parallel to the polarization of the excitation beam.

References 1. A B Djurišić, A M C Ng, X Y Chen, Prog. Quant. Electron., 34, 191 (2010). 2. D. Comedi, M. Tirado, C. Zapata, S.P. Heluani, M. Villafuerte, P. Mohseni, R.R. LaPierre, J. Alloys Compd., 495, 439 (2010). 3. A. B. Djurišić and Yu Hang Leung, Small, 2, No. 8-9, 944 – 961 (2006). 4. Min Gao, Rui Cheng, Wenliang Li, Yanping Li, Xiaoxian Zhang and Sishen Xie, J. Phys. Chem. C, 114, No. 25, (2010). 5. S. L. Chen, W. M. Chen, and I. A. Buyanova, Phys. Rev. B, 83, 245212 (2011).

72 Downloaded 27 Apr 2012 to 93.206.60.180. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp