Open Access
Add to BibSonomyAbstract
The optical transitions of the three-dimensionally confined GaN/AlGaN asymmetric multi quantum disks were characterized by micro photoluminescence and time-resolved photoluminescence. Several fine emission lines, originating from the wide and narrow quantum disks, were observed around 3.7 eV from a single nanocolumn dispersed on a patterned SiO2 substrate. The photoluminescence from the wide quantum disk shifts a little with increasing excitation power, while that from the narrow quantum disk does not shift. This effect can be explained by carrier tunneling for the 3-dimensionally confined quantum disks. Kelvin probe force microscopy results confirm that the GaN/AlGaN multiquantum disks are surrounded by a GaN shell, which has a higher potential than core GaN.
© 2015 Optical Society of America
1. Introduction
III–nitride semiconductor materials, such as GaN, InN, and AlN, have attracted much attention for the fabrication of high-brightness light emitting diodes and laser diodes [1,2]. It is well known that wurzite III-nitride structures with (0001) surface orientation have pyroelectric and built-in piezoelectric properties leading to strong electric fields along the crystal c-axis. This results in tilting of the electronic band structures at heterostructure interfaces [3,4] giving rise to the quantum confined Stark effect (QCSE) in quantum confined nanostructures. These strong electric fields produce a shift in emission line energies and increase the spatial separation of electron- and hole- wavefunctions in quantum wells, consequently reducing the quantum efficiency of III-nitride devices. However, nanocolumn heterostructures including quantum disks (QDisks) grown at the tip of GaN nanocolumns can produce new effects arising from the lateral quantum confinement, and strain relaxation. The novelty in these structures is twofold: on the one hand, the material system and the nanoscale dimensions in a columnar geometry. On the other hand, the relationship between strain, piezoelectric field, and potential confinement offered in such heterostructures gives rise to interesting physical phenomena which can be exploited in future applications. First steps have indeed been made towards practical devices with GaN nanocolumns, experiments employing Bragg reflectors, electrical excitation and stimulated emission having been undertaken showing encouraging results [5–7]. Lee and associates reported an enhanced carrier screening due to the tunnelling in coupled GaN/AlGaN asymmetric multiple quantum disks (AQDisks) consisting of a wide and a narrow QDisk (WQdisk and NQDisk) separated by an AlGaN barrier [8,9]. Such structures have previously been studied extensively in III-As systems to investigate tunnelling properties [10]. Advances in the field have led to the successful demonstration of photovoltaic devices and single wire optically pumped lasers in core-shell InGaN/GaN structures [11,12]. In recent reports, Rigutti et al have demonstrated a UV photodetector based on GaN/AlN multi quantum disks (MQDisks) [13] and Carnevale et al demonstrated the coaxial GaN/AlN superlattice by dynamically adjusting the growth kinetics [14].
In this study, 3-dimensionally confined coupled GaN/AlGaN AQDisks were grown systematically on the tip of GaN nanorods by using plasma-assisted molecular beam epitaxy (PAMBE). The potentials in the GaN shell and the core GaN QDisks are found to differ, with the shell being higher in potential than the core despite the material being identical. This potential difference was verified by measurements with Kelvin probe force microscopy (KPFM). Optical transitions demonstrating processes such as the quantum confined Stark effect and carrier tunnelling from the AQDisks were characterized by micro-photoluminescence (μPL) and time-resolved photoluminescence (TRPL).
2. Experimental
Self-assembled GaN/AlGaN MQDisks were grown at the tip of the GaN nanorods. The GaN nanorods were grown on a Si (111) substrate by using PAMBE with no buffer layer. Detailed growth conditions can be found in references [15–17]. Two Qdisk samples were prepared for this study as shown in Fig. 1.For MQDisks, one was made up of 10 periods of GaN/AlGaN QDisks with GaN well and AlGaN barrier thicknesses of ~4 and ~6 nm, respectively. For AQDisks, the other comprised two alternating GaN Qdisk of thickness ~1.5 nm and ~2.5 nm separated by an Al0.5Ga0.5N barrier with an effective thickness of ~2 nm, producing asymmetrically coupled quantum disks. Here, the larger Q-disk is referred to as DA, while the smaller Qdisk of GaN is referred to as DB. These MQDisks were surrounded by GaN not only in the vertical direction but also in the lateral direction. Details of the core-shell structures were confirmed by transmission electron microscopy and AFM analysis.
Fig. 1 (a) SEM top view image of the GaN/AlGaN AQdisks grown on GaN nanorods. (b) Schematic diagram of the sample structures.
KPFM measurements were performed in non-contact mode using a Bruker-Nano N8 Neos AFM system. For the measurements, we used conductive Si tips with Pt coating and a radius of less than 25 nm. The topography and surface potential (contact potential difference) of the GaN nanorods were measured simultaneously by using the amplitude modulation technique with two different frequencies. For the topography measurement the tip oscillated at the resonance frequency of the cantilever of 75 kHz. For the surface potential measurements, an AC modulation voltage of ~2 V at a frequency of 30 kHz was used, superimposing upon a DC bias voltage applied between the sample and the grounded tip. The resulting electrostatic forces acting on the tip were detected by a lock-in technique analyzing local differences in the tip oscillation amplitude. The amplitude signal at the frequency used includes the DC bias voltage and the work function difference between the tip and the sample, which is the contact potential difference (VCPD) between the tip and sample surface, VCPD = (ϕtip - ϕsample)/-e. Here, ϕtip and ϕsample are the work function of the tip and the sample, respectively, and e is the elementary charge. The feedback circuit in the AFM system controls the DC bias voltage to compensate VCPD to make the amplitude signal null. Thus, the compensated DC voltage signal corresponds to the local surface potential of the sample surface. The resulting KPFM image maps the variation of surface potential corresponding to the relative work function on the surface with respect to tip work function.
A 325 nm He–Cd laser was used as an excitation light source for the macro-PL measurement. A frequency-tripled femtosecond Ti:sapphire laser (100 fs pulses at 76 MHz) operating at 266 nm was used for the excitation of the Q-disks for the μ-PL experiments. A 36 × reflecting objective was mounted on a piezo-stage held above the cryostat to both focus the incident laser beam to a spot size of ~0.8 μm2 and to collect the resulting luminescence in a confocal geometry. The luminescence was then dispersed by a 1200 l/mm reflective grating in a 0.3m spectrometer giving a spectral resolution of ~700 μeV and a spatial resolution of 0.8 μm. The signal was detected using a cooled charge coupled device (CCD) detector. TRPL measurements were carried out using the same experimental set up as above the dispersed PL was directed towards a photomultiplier connected to a commercial photon counting system (Becker&Hickl SPC-134), with a time resolution of ~130 ps. Measurements of the lifetimes of the confined states were then carried out over a range of excitation power densities.
3. Results and discussion
A scanning electron microscopy (SEM) image and schematic diagrams of the GaN/AlGaN AQDisks grown on the GaN nanocolumns are shown in Fig. 1. It can be seen from the SEM image in Fig. 1a that two types of nanorods exist in the sample, as we reported earlier [15]. One comprises compact nanorods (dark regions) and the other protruded nanorod (bright regions). These nanostructures were vertically aligned along the (0001) direction. The density and diameters of the GaN nanorods were ~109 cm−2 and 150~200 nm, respectively.
Figure 2(a) presents a transmission electron microscopy (TEM; JEOL-2000EX) image of the AQDisks. Three different regions can be observed clearly. A 10 period stack of GaN/AlGaN MQDisks was inserted into the middle of initial GaN nanorod and the final GaN nanorod layer as shown schematically in Fig. 1(b). Note that the shell surrounding the GaN/AlGaN MQDisks was also observed. In order to see the shell in more detail, the red highlighted region in Fig. 2(a) was enlarged in Fig. 2(b). Lattice fringes were clearly observed, indicating the single crystal nature of the nanorods. The TEM image was divided into three parts. Figures 2(c)-2(e) show zoomed images of the outlined regions as A, B, and C in Fig. 2(b) by a digital diffractogram method obtained from a fast Fourier transform (FFT). Each region corresponds to pure GaN, side shell, and MQDisks, respectively.
Fig. 2 (a) TEM image of the GaN/AlGaN MQDdisks. (b) The enlarged image of the red highlighted region in a. (c)~(e) zoomed images of the outlined regions as A, B, and C.
Macro-PL spectra taken at 10 K from of the AQDisks, MQDisks and the GaN nanocolumns with comparable height and outer diameter are shown in Fig. 3(a).A typical emission line with a full width at half maximum (FWHM) of less than 10 meV from the pure GaN nanorods is observed at 3.47 eV, corresponding to donor bound excitons [15]. However, the FWHM of the emission lines from the QDisk samples is very broad due to the ensemble of the multiple disks and their size distribution. In addition, the emission lines are observed at an energy above that of the GaN nanorods apart from a shoulder near 3.47 eV, implying that the emission originates from quantum confined states. In the case of MQDisks, broad emission was observed at 3.52 eV originating from the quantum disk of 6 nm width. In AQDisks however, the PL emission is broader than the MQDisks, and consists of two distinct emission peaks. The dominant peak is at 3.54 eV and a shoulder can be seen 3.58 eV coming from the wide and narrow quantum disks, respectively. It is very difficult to define the exact origin of the emission since the macro-PL consists of an ensemble of the many different sample dimensions.
Fig. 3 (a) Macro-PL spectra of bare GaN nanocolumns and of GaN/AlGaN MQDs grown on GaN nanocolumns taken at 10 K. (b) Excitation power dependent micro-PL spectra of the GaN/AlGaN ACMQDs. The excitation powers used for the black, green, and pink are 4, 200, and 720 kW/cm2, respectively. The inset depicts an SEM image of a single nanocolumn dispersed on a SiO2 substrate patterned with a metal reference grid. The scale bar of the enlarged image is 1 μm. (c) A μ-PL spectrum taken from a MQDs dispersed on SiO2 substrate as shown in (a) at an excitation power of ~200 kW/cm2.
In order to characterize the PL from a single nanocolumn, many nanocolumns consisting of AQDisks were mechanically removed from their growth substrate and dispersed over a SiO2 wafer patterned with a metal reference grid as shown in the inset in Fig. 3(b). White light can be used to focus the microscope on the reference marks and subsequently the laser spot can be positioned on the red squared region by moving a piezo-electric stage to study the individual nanocolumn structure. Figure 3(b) shows the excitation power-dependent μ-PL spectra of the AQDisks taken from a single nanocolumn. Differences compared to the macro-PL measurements can be seen clearly. Several fine emission lines can be observed with a blueshifted emission compared to the macro-PL data. At low excitation power (4 kW/cm2) two main peaks around 3.67 eV (denoted as DA) and 3.74 eV (denoted as DB) are observed. As the excitation power is increased, these two lines begin to dominate the spectra. Furthermore, the intensity of the DB peak increases more slowly relative to that of the DA, indicating that the origins of these two emission lines are clearly different. This observation is consistent with earlier studies involving GaAs/AlGaAs asymmetric QWs, which demonstrated nonlinear PL dependence with excitation power, due to carrier tunneling [18]. Vertically confined AlGaN/GaN asymmetric quantum disks have also been found to display this phenomenon due to carrier tunneling from the narrow QDisk to the wide one [9]. Here, the emissions of the DA and DB peaks correspond to transitions from a wide and a narrower QDisk, respectively. In addition, the peak energy of DB does not shift with excitation power, while the DA peak red-shifts slightly by 5 meV. We deduce that the dominant sharp emission lines are not affected by the QCSE as this would result in a blue shift with increasing excitation power as the increased carrier density screens the internal field. For comparison, the micro-PL spectrum of a different sample comprising nanorods with simple MQDisks of GaN(4 nm)/AlGaN(6 nm) dispersed on a SiO2 substrate, was shown in the lower panel of Fig. 3b. The D0X arising from bulk GaN is broad due to defects near the bottom of the GaN nanorod. A strong and sharp emission near 3.64 eV was observed arising from the GaN/AlGaN quantum confined states, where the emission energy is lower than that of AQDisks because the disks are wider here. One interesting point to note is that the confined state emission for this particular nanorod appears in the high-energy tail of the macro-PL. As we can see from the macro-PL, emission from the D0X of the bulk GaN and from the confined state can’t be resolved due to ensemble effects, while the micro-PL from a dispersed single nanocolumn clearly resolves these emissions.
Figure 4 shows μ-TRPL traces collected from peaks DA and DB at two different excitation powers. The TRPL traces show a single exponential decay identified with free carrier recombination, with fitted decay times of 624 ps for the 3.67 eV peak and 588 ps for the 3.74 eV peak at low excitation power density (4 kW/cm2), respectively. For an asymmetric double quantum well under flat-band conditions with thick and high barriers, under weak coupling, the two QWs can be regarded as being decoupled such that the eigen-states in each well can be regarded as independent [10]. Accordingly, the measured decay time of the DA peak is longer than that for DB. However, the quantum wells will be tilted due to the strain between GaN and AlGaN in the real states. Then the electron will be more localized in the wide quantum disks, leading to the longer decay times. These values are about twice as large as that for GaN/AlGaN AQDisks with only vertical confinement [9], and GaN/AlN quantum dots in nanowires [19,20]. Further evidence for this tunneling phenomenon can be deduced from the longer risetime seen for the DA emission, indicating that the carrier population in this state builds up by transfer from a higher energy state. With the increasing excitation power, the decay time of the DB peak is almost unchanged, while that of DA peak increases by about 100 ps. This matches with the result of excitation power dependent PL of the DA peak as shown in Fig. 3(b), in that lower photon energy generally is associated with a longer decay time. An explanation of the peak shift phenomena observed in the PL of AQDisks on variation of excitation power can be given if we consider tunneling effects. The QDisk giving rise to peak DA does not undergo carrier field screening. Instead, as the excitation power increases, carriers tunnel from the narrow QDisk to the wider QDisk, causing a redshift and increasing decay time with excitation power.
Fig. 4 Time-resolved PL spectra of the 3.67 eV (DA) and 3.75 eV (DB) of the asymmetric multiquantum disks at two different excitation power densities of 4 kW/cm2 and 600 kW/cm2 taken at 4.2 K.
In order to elucidate these μPL and μTRPL phenomenon, we suggest a schematic band diagram as shown in Fig. 5.Wurzite III-nitride heterostructures have a built-in piezoelectric field which tilts the bands, resulting in the spatial separation of the confined electrons and holes. In our calculations [9], the electron energy of DA lies on the triangular potential well, whereas that of DB lies above the triangular potential well, resulting in that emission energy and decay time of the DA are lower and longer than those of DB, respectively. Though the band bending occurs both disks, the energy of the DB is the same because the electron energy of DB lies above the triangular potential well. However, the energy of DA shifts and lifetime changes with excitation power. Red-shift and increased decay time of the DA emission can be given by considering the carrier’s tunneling effects from the narrow QDisk to the wider QDisk with increasing excitation power [9]. Further evidence for the tunneling phenomenon can be deduced from the longer rise time seen for the DA emission, indicating that the carrier population in this state builds up by transfer from a higher energy state. As power is increased the measured decay time of the DA peak is longer than that for the DB peak due to the additional band bending caused by electron tunneling from the narrow QDisks. The enhanced band bending makes the electron and hole in the wide QDisks more localized and separated, resulting in the longer decay times. If we consider that the well thicknesses are not too different each other, the tunneling rate can be estimated by 1/τnw = 1/τww + 1/τt, where τnw, τww, and τt are the narrow well, wide well lifetimes and the tunneling time, respectively [10]. Then the tunneling rate is estimated to be 2.74 ns.
Fig. 5 Schematic band diagram of the ACMQDs. (a) low excitation power and (b) high excitation power.
In order to determine the surface potential (or surface work function) distribution across the nanocolumns, atomic force microscopy (AFM) and Kelvin probe force microscopy (KPFM) measurements were undertaken. Figure 6(a) shows an AFM topographic image of the GaN nanocolumns, which is in good agreement with a SEM image showing that most of the GaN nanocolumns have hexagonal geometry. The distorted shape in the AFM image is attributed to tip artifacts due to the large difference of the nanocolumns heights in the topography measurement. The brighter nanorod has longer length because the brightness in the topographic image indicates the height.
Fig. 6 (a) and (b) present an atomic force microscopy image and Kelvin probe force microscopy image of the asymmetric multiquantum wells measured on the tip of the nanocolumns, respectively. The scale bars are 250 nm. (c) Potential line profile of the nanocolumn indicated in (b), which illustrates in more detail the potential distribution.
Figure 6(b) shows a KPFM image of the potential variation of the top surface on the GaN nanocolumns. The brightness variation in the KPFM image reflects the local surface work function difference with respect to the tip work function. When compared to the topography image, the brightness of the surface area of the individual nanocolumns varies from the core to the shell regions. The shell regions are darker than the core regions with a thickness of ~30 nm. There is also a non-uniformity in the brightness of the core regions. The brighter region has a smaller VCPD (contact potential difference) between the Pt-coated tip and the GaN nanorods [(ϕpt - ϕNR)/e], indicating relatively lower work function than other regions. Thus, the shell regions (darker regions) have a higher work function than the core regions (brighter regions). A difference in potential of ~0.5 eV was determined from the line profile of the KPFM image, as shown in Fig. 5c. This potential variation using KPFM was well explained by the authors of reference [21].
Possible explanations for the difference in the work function between shell and core regions are outlined below by considering the structure of GaN nanorods containing AlGaN quantum disks. In semiconductors, the work function is determined by electron affinity, band gap, and doping level. One possibility is the defect generation in the core regions during GaN overgrowth. If the relatively defect-free crystalline GaN layer is grown in the shell regions and imperfect GaN with several defects is grown in the core regions, the upward Fermi level shift in the core regions can generate a relatively smaller work function. Another possibility is the electron affinity variation due to the crystalline orientation of the GaN. As shown in Fig. 2b, the plane orientation of the shell regions is different from that of the core regions. Thus, the electron affinity of the shell regions is larger than that of the core regions, which can generate a higher work function in the shell regions [22,23].
4. Conclusion
In conclusion, we have characterized GaN/AlGaN asymmetric multi quantum disks optically and found a reduced Stark shift owing to three-dimensional confinement. Single nanocolumn structures were characterized and several fine emissions lines were observed around 3.7 eV. As the excitation power density is increased, the emission peak photon energy and decay time from a narrow quantum disk are almost unchanged, while those from a wide quantum disk undergo a slight redshift and increase in decay time, in contrast to systems involving strained III-nitride QWs. The decay times of these systems are about twice those seen in vertically confined structures. These effects can be explained by carrier tunneling for the 3-dimensionally confined quantum disks. The KPFM results confirm that the GaN/AlGaN multi-quantum disks are surrounded by a GaN shell, which has a higher potential than core GaN.
Acknowledgments
This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2013R1A1A2011426)
References and links
1. F. A. Ponce and D. P. Bour, “Nitride-based semiconductors for blue and green light-emitting devices,” Nature 386(00), 351–359 (1997). [CrossRef]
2. S. Nakamura, M. Senoh, N. Iwasa, and S. I. Nagahama, “High-brightness InGaN blue, green and yellow light-emitting diodes with quantum well structures,” Jpn. J. Appl. Phys. 34(27A), L797–L799 (1995). [CrossRef]
3. A. Bykhovski, B. Gelmont, and M. Shur, “The influence of the strain induced electric field on the charge distribution in GaN-AlN-GaN structure,” J. Appl. Phys. 74(11), 6734–6739 (1993). [CrossRef]
4. F. Bernardini, V. Fiorentini, and D. Vanderbilt, “Spontaneous polarization and piezoelectric constants of III-V nitrides,” Phys. Rev. B 56(16), R10024 (1997). [CrossRef]
5. J. Ristić, E. Calleja, A. Trampert, S. Fernández-Garrido, C. Rivera, U. Jahn, and K. H. Ploog, “Columnar AlGaN/GaN Nanocavities with AlN/GaN bragg reflectors grown by molecular beam epitaxy on Si(111),” Phys. Rev. Lett. 94(14), 146102 (2005). [CrossRef] [PubMed]
6. A. Kikuchi, M. Kawai, M. Tada, and K. Kishino, “InGaN/GaN multiple quantum disk nanocolumn light-emitting diodes grown on (111) Si substrate,” Jpn. J. Appl. Phys. 43(12A12A), L1524–L1526 (2004). [CrossRef]
7. A. Kikuchi, K. Yamano, M. Tada, and K. Kishino, “Stimulated emission from GaN nanocolumns,” Phys. Status Solidi B 241(12), 2754–2758 (2004). [CrossRef]
8. K. H. Lee, S. Birner, J. H. Na, R. A. Taylor, S. N. Yi, Y. S. Park, C. M. Park, and T. W. Kang, “Two-photon excitation spectroscopy of coupled asymmetric GaN/AlGaN quantum discs,” Nanotechnology 17(23), 5754–5758 (2006). [CrossRef]
9. K. H. Lee, J. H. Na, R. A. Taylor, S. N. Yi, S. Birner, Y. S. Park, C. M. Park, and T. W. Kang, “Enhancement of free-carrier screening due to tunneling in coupled asymmetric GaN/AlGaN quantum discs,” Appl. Phys. Lett. 89(2), 023103 (2006). [CrossRef]
10. J. Shah, Ultrafast Spectroscopy of Semiconductors and Semiconductor Nanostructures (Springer, 1996).
11. Y. Dong, B. Tian, T. J. Kempa, and C. M. Lieber, “Coaxial group III-nitride nanowire photovoltaics,” Nano Lett. 9(5), 2183–2187 (2009). [CrossRef] [PubMed]
12. S. Gradecak, F. Qian, Y. Li, H.-G. Park, and C. M. Lieber, “GaN nanowire lasers with low lasing thresholds,” Appl. Phys. Lett. 87(17), 173111 (2005). [CrossRef]
13. L. Rigutti, M. Tchernycheva, A. De Luna Bugallo, G. Jacopin, F. H. Julien, L. F. Zagonel, K. March, O. Stephan, M. Kociak, and R. Songmuang, “Ultraviolet photodetector based on GaN/AlN quantum disks in a single nanowire,” Nano Lett. 10(8), 2939–2943 (2010). [CrossRef] [PubMed]
14. S. D. Carnevale, J. Yang, P. J. Phillips, M. J. Mills, and R. C. Myers, “Three-dimensional GaN/AlN nanowire heterostructures by separating nucleation and growth processes,” Nano Lett. 11(2), 866–871 (2011). [CrossRef] [PubMed]
15. Y. S. Park, C. M. Park, D. J. Fu, T. W. Kang, and J. E. Oh, “Photoluminescence studies of GaN nanorods on Si (111) substrates grown by molecular-beam epitaxy,” Appl. Phys. Lett. 85(23), 5718–5720 (2004). [CrossRef]
16. C. M. Park, Y. S. Park, H. Im, and T. W. Kang, “Optical properties of GaN nanorods grown by molecular-beam epitaxy; dependence on growth time,” Nanotechnology 17(4), 952–955 (2006). [CrossRef] [PubMed]
17. Y. S. Park, S. H. Lee, J. E. Oh, C. M. Park, and T. W. Kang, “Self-assembled GaN nano-rods grown directly on (111) Si substrates: Dependence on growth conditions,” J. Cryst. Growth 282(3–4), 313–319 (2005). [CrossRef]
18. D. J. Leopold and M. M. Leopold, “Tunneling-induced optical nonlinearities in asymmetric Al0.3Ga0.7As/GaAs double-quantum-well structures,” Phys. Rev. B Condens. Matter 42(17), 11147–11158 (1990). [CrossRef] [PubMed]
19. J. Renard, R. Songmuang, G. Tourbot, C. Bougerol, B. Daudin, and B. Gayral, “Evidence for quantum-confined Stark effect in GaN/AlN quantum dots in nanowires,” Phys. Rev. B 80(12), 121305 (2009). [CrossRef]
20. K. Choi, S. Kako, M. J. Holmes, M. Arita, and Y. Arakawa, “Strong exciton confinement in site-controlled GaN quantum dots embedded in nanowires,” Appl. Phys. Lett. 103(17), 171907 (2013). [CrossRef]
21. S. Kurbanov, W. C. Yang, and T. W. Kang, “Kelvin probe force microscopy of defects in ZnO nanocrystals associated with emission at 3.31 eV,” Appl. Phys. Express 4(2), 002110 (2011). [CrossRef]
22. L. Diederich, O. M. Kuttel, P. Aebi, and L. Schlapbach, “Electron affinity and work function of differently oriented and doped diamond surfaces determined by photoelectron spectroscopy,” Surf. Sci. 418(1), 219–239 (1998). [CrossRef]
23. M. Prutton, Surface Physics (Oxford Univ. Press, 1983) 2nd ed. P78.
