Study of the internal quantum efficiency of InGaN/GaN UV LEDs on patterned
sapphire substrate using the electroluminescence method
C.H. Wang
a, C.C. Ke
a, C.H. Chiu
a, J.C. Li
a, H.C. Kuo
a,n, T.C. Lu
a,b, S.C. Wang
aa
Department of Photonics and Institute of Electro-Optical Engineering, National Chiao Tung University, 1001 University Road, Hsinchu 30050, Taiwan
bInstitute of Lighting and Energy Photonics, National Chiao Tung University, 301 Gaofa 3rd Road, Guiren Township, Tainan County 711, Taiwan
a r t i c l e
i n f o
Available online 24 September 2010
Keywords: A1. Substrates
A3. Metalorganic chemical vapor deposition
B1. Nitrides
B3. Light emitting diodes
a b s t r a c t
Internal quantum efficiency (IQE) of InGaN/GaN UV LEDs with patterned sapphire substrates (PSS) was investigated using electroluminescence (EL) and photoluminescence (PL) methods. The physical mechanisms that affect temperature-dependent EL efficiency as a function of injected carrier density were deduced. In order to reduce the density of non-radiative recombination centers to improve quantum efficiency, improvement of crystal quality and reduction in the number of defects are necessary. PSS LEDs showed better EL characteristics than non-PSS LEDs because of the improved epitaxial quality. Contrary to PL IQE, EL IQE was significantly affected by carrier injection efficiency, especially at low temperature when high bias voltage was applied to the p–n junction.
Crown Copyright & 2010 Published by Elsevier B.V. All rights reserved.
1. Introduction
III-nitride based light emitting diodes (LEDs) have attracted much attention for green to ultraviolet (UV) wavelengths in the last decade due to the tunable wide band gap. Recently, LEDs in the UV range have been the focus of research because of their
tremendous potential in applications like white light sources[1],
environmental cleaning, and bio-medical examination. It is well known that high efficiency radiative recombination in blue and green InGaN multiple quantum wells LEDs has been mainly attributed to the localized states due to phase separation or
fluctuations in In mole fraction in InxGa1 xN/GaN quantum wells
[2]. For UV LEDs, InN mole fraction in the InGaN well is lower than
that for blue or green LEDs, so the radiative efficiency of such LEDs is more sensitive to the threading dislocation density.
Some research groups have investigated the temperature dependence of electroluminescence (EL) intensity of InGaN/GaN
QW diodes [3,4], which reveals anomalous EL quenching at
temperatures below 100 K. It was found that the anomalous temperature dependence of EL efficiency is caused by the interplay of radiative recombination in QWs and the carrier injection coefficient. One of the anticipated genuine causes for low temperature EL quenching may be ascribed to the deep Mg
acceptor level of 170 meV in p-GaN[5], which can be deactivated
at lower temperatures below 100 K. Therefore, holes fail to be injected into the QW active region from the p-GaN layer,
especially when an electron blocking p-type AlGaN barrier is introduced.
Moreover, although the external quantum efficiency (EQE) of UV LEDs grown on patterned sapphire substrate (PSS) can be significantly improved due to the low dislocation density and high
extraction efficiency (
Z
lee), the actual enhancement in IQE andZ
leeis still hard to quantify. In this study, we analyzed the detailed physical mechanisms for EL enhancement effects under forward bias conditions, and the effects of dislocation density in UV LEDs were also examined. EL behavior under low and high injection current is investigated from low temperature (LT) to room temperature (RT). The EL IQE results were compared to photo-luminescence (PL) IQE results to investigate the difference between these methods.
2. Experiments
The etched patterns on sapphire substrate are 2D arrays of holes
arranged in a hexagonal lattice with depth 1.4
m
m and diameter3.5
m
m, and the distance between each pattern is 1.0m
m. Thedetailed process can be found in Ref.[6]. The LED structure was then
grown on a planar substrate and PSS by low-pressure metal–organic chemical vapor deposition.
The epitaxial structure consisted of a 30-nm-thick AlN
nucleation layer, a 2-
m
m-thick Si-doped n-type GaN, and anunintentionally doped active layer with InxGa1 xN/GaN MQWs, a
20-nm-thick AlxGa1 xN electron blocking layer, and a 0.2-
m
m-thick Mg-doped p-type GaN. The doping concentrations of n- and
p-type GaN were nominally 5 1018 and 1 1019cm 3,
Contents lists available atScienceDirect
journal homepage:www.elsevier.com/locate/jcrysgro
Journal of Crystal Growth
0022-0248/$ - see front matter Crown Copyright & 2010 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2010.09.046
n
Corresponding author. Tel.: + 886 3 571 2121x31986; fax: +886 5 571 6631. E-mail address: [email protected] (H.C. Kuo).
respectively. The MQW comprised 16 periods of an In0.15Ga0.85N well ( 2 nm) and a GaN barrier ( 16 nm), and the emission wavelength of both samples is around 400 nm at 20 mA. The LEDs were fabricated through the standard process, and the size of LED
mesas was 300 300
m
m2.For the temperature-dependent EL measurement, the sample was mounted in a closed-cycle helium cryostat and temperature was controlled from 15 to 300 K. Output light was detected by a 0.32 m monochromator (Jobin-Yvon Triax-320) with 1800, 1200, and 300 grooves/mm gratings and the maximum width of the entrance slit was 1 mm. The injection current was from 0.1 to 80 mA, which can be calculated in terms of current density as
from 0.11 to 88 A/cm2. The details of the temperature-dependent
EL measurement setup are shown inFig. 1. PL was excited by a
femto-second-pulse Ti:Sapphire laser whose output wavelength could be tuned to be around 380–390 nm to avoid absorption by GaN and ensure that most of the photons generated by the laser light could excite the photons in the MQWs. Details of the PL IQE measurement and the definition of injected carrier density were
published in Ref.[7].
3. Results and discussion
The temperature-dependent relative IQE curves versus
injec-tion current are summarized inFig. 2(a) and (b), for the PSS and
non-PSS LEDs, respectively. We define the relative IQE as the output power collected from the LED divided by the injection
current. Generally, EQE can be defined as the product of
Z
leeandIQE. Since precise
Z
leeis hard to accurately measure and separatefrom IQE, we use normalization to eliminate
Z
lee, which is almostconstant under various temperatures and injection currents. For every specific condition, the maximum efficiency is set as 1 to normalize others. At first glance, one could find that at every temperature, as the injection current increased, there is a peak efficiency. However, this peak position varied as temperature changed. For example, it was around 8 mA when temperature was above 120 K and gradually moved to 1 mA when temperature decreased to 80 K. Moreover, the maximum in all the measure-ment curves happened at around 80 and 100 K for PSS LED and non-PSS LED, respectively, instead of the lowest temperature. To understand this phenomenon, further investigation is needed.
Fig. 3(a) and (b) shows the temperature-dependent relative IQE at 0.1 and 20 mA, respectively. It can clearly be seen that at the low injection current of 0.1 mA, relative EL efficiencies for
both PSS and non-PSS LEDs decrease monotonously with increase in temperature. But at the injection current of 20 mA, EL efficiency first increases and then decreases with increase in temperature.
Fig. 1. Schematic of electroluminescence measurement setup.
0.1 1 10 100 0.0 0.2 0.4 0.6 0.8 1.0 80K 140K 220K 300K Relative IQE (%) Current (mA) 0.1 1 10 100 0.0 0.2 0.4 0.6 0.8 1.0 80K 160K 220K 300K Relative IQE (%) Current (mA)
Fig. 2. Temperature-dependent relative IQE curves of (a) PSS and (b) non-PSS LEDs.
It is worth noting that EL efficiency for the PSS LED is much higher than that of non-PSS LED when temperature is above 150 K at both low and high injection currents. There are two possible mechanisms at play with increasing temperature. Firstly, in InGaN-based structures, emission comes from the localized states, which can trap carriers within their potential minima. When temperature increases, the carriers receive thermal energy and will be activated from potential minima to non-radiative centers. Therefore, EL efficiency will decrease. In the following, this is denoted as carrier delocalization effect. Secondly, it is well known that hole concentration and mobility decrease when temperature decreases, leading to increase in series resistance and forward
voltage[8]. That is to say, the external field applied on the LED
structure is much higher at low temperature, which will result in more carriers escaping out of the well region as well as a decrease in radiative recombination rate. In the following, this is denoted as external field effect.
To figure out which mechanism is dominant at different injection currents, we investigated the temperature-dependent
I–V curves, as shown inFig 4(a) and (b). At an injection current of
20 mA, the forward voltage increases by about 1 and 1.4 V from 300 to 77 K for PSS and non-PSS LEDs, respectively, while at an injection current of 0.1 mA, the variation in forward voltage is much smaller for both LEDs (only about 0.5 V). These results
indicate that the external field effect plays a more important role at high injection current. Therefore, we can summarize that at low injection current, the decrease in EL efficiency with increase in
0 50 100 150 200 250 300 0.0 0.2 0.4 0.6 0.8 1.0 Relative IQE (%) Temperature (K) PSS @ 0.1mA Non-PSS @ 0.1mA 0 50 100 150 200 250 300 0.0 0.2 0.4 0.6 0.8 1.0 PSS @ 20mA) Non-PSS @ 20mA Relative IQE (%) Temperature (K)
Fig. 3. Temperature-dependent relative IQE of PSS and non-PSS LEDs at (a) 0.1 mA and (b) 20 mA. 0 1 2 3 4 5 0 20 40 60 80 PSS 77 140 220 300 Current (mA) Voltage (V) 0 1 2 3 4 5 0 20 40 60 80 Non-PSS 77 140 220 300 Current (mA) Voltage (V)
Fig. 4. Temperature-dependent I–V curves of (a) PSS and (b) non-PSS LEDs.
1E12 1E13 1E14 1E15 1E16 1E17 1E18 0.0 0.2 0.4 0.6 0.8 1.0 LT RT EL IQE PSS @LT NonPSS @LT PSS @RT NonPSS @RT PL IQE PSS @LT NonPSS @LT PSS @RT NonPSS @RT Relative IQE (%) Carrier density (#/cm3)
Fig. 5. Relative electroluminescence and photoluminescence IQE as a function of injection current at low temperature and room temperature.
C.H. Wang et al. / Journal of Crystal Growth 315 (2011) 242–245 244
temperature is mainly due to the carrier delocalization effect. But at high injection current, due to the competition of carrier delocalization effect and external field effect, EL efficiency reaches a maximum value at higher temperature. On the other hand, as temperature was decreased from 300 to 77 K, the turn-on voltages of PSS and non-PSS LEDs were increased from 2.98 to 3.74 V and 3.11 to 4.02 V, respectively. This phenomenon indicates that carrier injection in the diodes is more difficult at low temperature.
Furthermore, we compared the relative EL IQE with PL IQE as a function of carrier density at low and room temperature as shown inFig. 5. In PL IQE measurements, since PL efficiency was almost constant for temperatures below 30 K, here we used 15 K as LT, and for EL IQE, we chose 77 K as the LT since the maximum relative IQE occurred at about 77 K. The definition of carrier density in EL measurement is the injected carrier divided by the active volume, while in PL measurement, we adopted the method
in Ref.[7].
The relative PL IQE is higher than the relative EL IQE at low carrier density. This phenomenon indicates that more carrier leakage happens when carriers are injected using a current source, which causes non-uniform distribution of holes. We found that EL efficiency decreased more rapidly at high injected carrier density. This can be attributed to the forward voltage increasing with increasing injected carrier density. Since the injected electrons inefficiently recombined with holes, they easily reached the p-GaN layer, resulting in EL efficiency drooping more rapidly at LT than RT. Efficiency at RT measured by the EL method is slightly lower than that measured by PL. These results imply that EL efficiency is significantly affected by carrier injection efficiency. To increase EL efficiency, a better LED chip or epitaxial structure design might be required.
In addition, the reduction of efficiency after the maximum in the curves is the well-known efficiency droop behavior, which can
be attributed to carrier overflow[9], non-uniform distribution of
holes [10,11], Auger scattering [12], and carrier delocalization
[13]. Comparing EL efficiency at RT with the maximum efficiency
at LT, the relative IQE of the PSS LED (59.3%) is much higher than that of the non-PSS LED (42.3%) at 20 mA. However, the reduction of relative IQE with increasing injection current is almost the same in both LEDs. These results imply that the reduction of dislocation density effectively improves the efficiency but does not affect the efficiency droop behavior at high injection current,
which is consistent with the results in Ref.[14].
4. Conclusion
From our analysis, we deduced the physical mechanisms that affect the temperature dependence of EL efficiency as a function of injected carrier density. The non-radiative recombination centers and the carrier overflow mechanisms both play important roles in the variation of the quantum efficiency. In order to reduce the density of non-radiative recombination centers to improve the quantum efficiency at low injected carrier density, improvement of crystal quality and reduction in density of defects are necessary. Indeed, an LED grown on PSS substrate shows higher efficiency than an LED grown on non-PSS substrate. Finally, we compared EL IQE with PL IQE. EL IQE was found to be smaller than PL IQE, which indicates that to equilibrate EL and PL IQE, carrier injection efficiency and the effect of temperature cannot be ignored.
Acknowledgments
The authors would like to thank Epistar Corporation for their technical support. This work was funded by the National Science Council in Taiwan under Grant number NSC98-3114-M-009-001. References
[1] Y. Narukawa, I. Niki, K. Izuno, M. Yamada, Y. Murazaki, T. Mukai, Jpn. J. Appl. Phys. 41 (2002) L371.
[2] Y. Narukawa, Y. Kawakami, M. Funato, Sz. Fujita, S. Fujita, S. Nakamura, Appl. Phys. Lett. 70 (1997) 981.
[3] A. Hori, D. Yasunaga, A. Satake, K. Fujiwara, Appl. Phys. Lett. 79 (2001) 3723. [4] C.L. Yang, L. Ding, J.N. Wang, K.K. Fung, W.K. Ge, H. Liang, L.S. Yu, Y.D. Qi,
D.L. Wang, Z.D. Lu, K.M. Lau, J. Appl. Phys. 98 (2005) 023703.
[5] S. Grzanka, G. Franssen, G. Targowski, K. Krowicki, T. Suski, R. Czemecki, P. Perlin, M. Leszczyski, Appl. Phys. Lett. 90 (2007) 103507.
[6] J.H. Lee, J.T. Oh, J.S. Park, J.W. Kim, Y.C. Kim, J.W. Lee, H.K. Cho, Phys. Status Solidi C 3 (2006) 2169.
[7] C.H. Ya-Ju Lee, P.C. Chiu, T.C. Lin, H.C. Lu, Kuo, S.C. Wang, IEEE J. Sel. Top. Quantum Electron. 15 (2009) 1137.
[8] X.A. Cao, S.F. LeBoeuf, IEEE Trans. Electron Devices 54 (2007) 3414. [9] K.J. Vampola, M. Iza, S. Keller, S.P. DenBaars, S. Nakamura, Appl. Phys. Lett. 94
(2009) 061116.
[10] K. Ding, Y.P. Zeng, X.C. Wei, Z.C. Li, J.X. Wang, H.X. Lu, P.P. Cong, X.Y. Yi, G.H. Wang, J.M. Li, Appl. Phys. B 97 (2009) 465.
[11] C.H. Wang, J.R. Chen, C.H. Chiu, H.C. Kuo, Y.L. Li, T.C. Lu,and, S.C. Wang, IEEE Photon. Technol. Lett. 22 (2010) 236.
[12] A. David, M.J. Grundmann, Appl. Phys. Lett. 96 (2010) 103504. [13] B. Monemar, B.E. Sernelius, Appl. Phys. Lett. 91 (2007) 181103.
[14] A. Hori, D. Yasunaga, A. Satake, K. Fujiwara, Appl. Phys. Lett. 91 (2007) 231114.
