GaN/ZnO and AlGaN/ZnO heterostructure LEDs: growth, fabrication, optical and electrical characterization

Transcription

1 Mater. Res. Soc. Symp. Proc. Vol Materials Research Society 1201-H01-08 GaN/ZnO and AlGaN/ZnO heterostructure LEDs: growth, fabrication, optical and electrical characterization J. Benz1, S. Eisermann1, P. J. Klar1, B. K. Meyer1, T. Detchprohm2 and C. Wetzel2 I. Physikalisches Instiut, Justus-Liebig Universität, Heinrich-Buff-Ring 16, Giessen, Germany 2 Department of Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic Institute, 110 Eighth Street, Troy, NY , U.S.A. 1 ABSTRACT The wide bandgap polar semiconductors GaN and ZnO and their related alloys exhibit fascinating properties in terms of bandgap engineering, carrier confinement, internal polarisation fields, and surface terminations. With a small lattice mismatch of ~1.8 % between GaN and ZnO and the possibility to grow MgZnO lattice-matched to GaN, the system AlGaN/MgZnO offers the opportunity to design novel optoelectronic devices circumventing the problem of p-type doping of ZnO. In such AlGaN/MgZnO heterostructures with either hetero- or isovalent interfaces, tuning of band offsets is possible in various ways by polarisation fields, surface termination, strain, and composition. These aspects need to be fully understood to be able to make full use of this class of heterostructures. We report on the growth of ZnO films by chemical vapor deposition on p-type GaN and AlGaN grown by metal-organic vapor deposition on sapphire templates and on the fabrication of corresponding light-emitting diode (LED) structures. Electrical and optical properties of the n-zno/p-gan and n-zno/p-algan LEDs will be compared and the observed differences will be discussed in terms of the band alignment at the heterointerface. INTRODUCTION The wide bandgap semiconductor zinc oxide is a promising material for the production of blue and ultraviolet optoelectronic devices such as light emitting diodes (LEDs), laser diodes and photo diodes. Compared to GaN, the semiconductor mainly used in the optoelectronic industry for the production of short wavelength devices; ZnO offers several advantages, e.g. larger exciton binding energy (60 mev for ZnO versus 26 mev for GaN), which may lead to UV sources with higher brightness and lower power thresholds at room temperature. ZnO possesses higher radiation hardness than Si, GaAs, CdS and GaN, therefore it should be suitable for space applications. Last but not least, large area substrates of ZnO are available at relatively low material costs and ZnO offers a simplified processing, as it can be microstructured by conventional wet-chemical etching [1-4]. Despite all these advantages, there remains one obstacle to be overcome before reliable, entirely ZnO-based optoelectronic devices become reality: the problem of p-type doping of ZnO. So far, there is no way to reliably produce stable and high quality p-type ZnO. Considering the similarity of the physical properties of ZnO and GaN, which both crystallize in wurtzite structure with a lattice mismatch of ~1.8%, and the availability of high quality p-gan, a natural way of circumventing the doping issue is to grow nmgzno/p-algan based devices. Several groups have started to work in this field growing n-zno/p-gan heterostructures. Such heterostructure devices are supposed to exhibit improved

2 current confinement compared to homojunctions, which could lead to higher recombination rates at the interface and thus higher device efficiency [1-4]. These heterostructures must be considered a first step towards light-emitting devices with more complicated active regions consisting of MgZnO/ZnO quantum wells. EXPERIMENT Schematic diagrams of the two heterostructures investigated are shown in figure 1. The growth of the GaN templates was done at Rensselaer Polytechnic Institute. The templates were grown by metal-organic vapor-phase epitaxy (MOCVD). A 3.7 µm thick GaN buffer layer was deposited on a sapphire substrate, followed by a layer of weakly Mg doped GaN of 325 nm thickness and a highly doped GaN:Mg, AlGaN:Mg layer of 25 nm thickness, respectively. The unintentionally doped n-type ZnO layer was grown on to these templates by chemical vapor deposition (CVD) in Giessen. The hole density of the GaN templates was determined to be p~1018 cm-3 at room temperature and for the electron density of the ZnO layer a rough approximation is n~1017 cm-3. Photolithography followed by a wet-chemical etching step in a solution of water (H2O), phosphoric acid (H3PO4) and acetic acid (C2H4O2) in parts of 30:1:1 by volume was used for the fabrication of mesa structures. The square shaped ZnO mesas have edge lengths of about 1200 µm. The contact areas on top of the mesa and surrounding the mesa were fabricated by evaporating Au on thin layers of Cr and Ni to improve adhesion, respectively. The inset in figure 2 shows a contacted sample. Figure 1. Schematic diagram of the ZnO/GaN and ZnO/AlGaN heterostructure. DISCUSSION The measured I-V characteristics at room temperature presented in figure 2 show a clearly rectifying, diode-like behavior with threshold voltages of ~3.4 V for ZnO/GaN and ~4.8 V for ZnO/AlGaN, respectively. The reverse breakdown voltages are estimated to be ~8.6 V for the ZnO/GaN structure and ~12.1 V for the ZnO/AlGaN structure, respectively. The ideality factor n and the saturation current Is are calculated from a modified diode equation (see equation 1), which takes the series resistance Rs of the layers and contacts into account.

3 U=(nkT/e)ln(I/Is)+RsI (1) The ideality factor is determined to be n = 26, the saturation current to Is = 4.7 x 10-7 A and the series resistance to Rs = 11 kω for ZnO/GaN and to n = 36, Is = 2.6 x 10-7 A and Rs = 7 kω for ZnO/AlGaN. We assume that these high values for n arise from the interplay of multiple transport mechanisms at the heterointerface. The comparatively high values for the series resistances and for the threshold voltages compared to the literature [2,4] indicate that the contacts need to be improved. Figure 2. I-V characteristics of the ZnO/GaN and ZnO/AlGaN heterostructure. The inset shows a contacted sample. Figure 3. Integral EL of a ZnO/AlGaN sample as a function of (a) reverse and (b) forward bias. The inset shows the bright green EL obtained under reverse bias.

4 The integral electroluminescence (EL) was measured by placing the sample in front of a photomultiplier. Figure 3 shows the intensity as a function of the applied current. We obtained a linear relationship under reverse bias, corresponding to recombination of free carriers at localized defects and impurities. Under forward bias the relationship is parabolic, which we attribute to radiative band to band emission in the interface region. The intensity under reverse bias is one order of magnitude higher than that under forward bias, thus we were not able to measure the spectral distribution of the intensity under forward bias, while the intensity under reverse bias was bright enough to be seen with the naked eye in a dark room and therefore could be further analyzed spectrally. Figure 4 shows the spectral EL of a ZnO/AlGaN sample under a reverse bias of 15 V and obtained using a UV spectrometer and lock-in technique. The spectrum consists of a broad peak centered at 550 nm. The band may originate from defect levels, probably Cu, in ZnO which act as efficient radiative traps for the minority holes. Alternatively it may also be due to deep defect levels in GaN [5]. The inset shows the I-V characteristics before and after applying a reverse bias of 15 V for several hours to measure the EL, the value for the reverse breakdown voltage decreases significantly indicating that the device structure has changed under high DC bias. Figure 4. Spectral EL of a ZnO/AlGaN sample. The inset shows the I-V characteristics before and after applying a reverse bias of 15 V for several hours to measure the EL. CONCLUSIONS Here, we reported on the fabrication of n-zno/p-gan and n-zno/p-algan heterostructure diodes. We investigated the current-voltage characteristics and achieved a clear rectifying behavior. We found a weak electroluminescence (EL) under forward bias as well as reverse bias for both heterosystems. The EL under reverse bias very likely originates from defect-related minority carrier recombination in ZnO or GaN, whereas the weaker EL under forward bias is

5 partly due to band edge emission. Further improvement is anticipated by including MgZnO/ZnO quantum wells in the active region. REFERENCES 1. D. C. Look, Mater. Sci. Eng., B 80, 383 (2001) 2. Y. I. Alivov, J. E. Van Nostrand, D. C. Look, M. V. Chukichev, B. M. Ataev, Appl. Phys. Lett., 83, 2943 (2003) 3. D. J. Rogers, F. Hosseini Teherani, A. Yasan, K. Minder, P. Kung, M Razeghi, Appl. Phys. Lett., 88, (2006) 4. Y. I. Alivov, E. V. Kalinina, A. E. Cherenkov, D. C. Look, B. M. Ataev, A. K. Omaev, M. V. Chukichev, D. M. Bagnall, Appl. Phys. Lett., 83, 4719 (2003) 5. I. E. Titkov, A. S. Zubrilov, L. A. Delimova, D. V. Mashovets, I. A. Liniichuk, I. V. Grekhov, Semiconductors, 41, 564 (2007)