Nanoscale View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
PAPER
View Journal | View Issue
Cite this: Nanoscale, 2014, 6, 4371
Light-extraction enhancement of a GaN-based LED covered with ZnO nanorod arrays† Hyun Jeong,a Doo Jae Park,a Hong Seok Lee,d Yeong Hwan Ko,e Jae Su Yu,e Sang-Bae Choi,f Dong-Seon Lee,f Eun-Kyung Suh*c and Mun Seok Jeong*ab We investigate the mechanism of light extraction enhancement of a GaN-based light-emitting diode (LED) grown on patterned sapphire substrate (PSS), that has ZnO nanorod arrays (NRAs) fabricated on top of the device using the hydrothermal method. We found that the light output power of the LED with ZnO NRAs increases by approximately 30% compared to the conventional LED without damaging the electrical properties of the device. We argue that the gradual decrease of the effective refractive index, which is caused by the fabrication of ZnO NRAs, is the mechanism of the observed improvement. Our argument is confirmed by cross-sectional confocal scanning electroluminescence microscopy (CSEM) and the
Received 12th December 2013 Accepted 19th January 2014
theoretical simulations, where we observed a distinct increase of the transmission at the interface between LED and air at the operation wavelength of the LED. In addition, the plane-view CSEM results indicate that ZnO NRAs, which were grown on the bare p-type GaN layer as an electrical safety margin
DOI: 10.1039/c3nr06584g
area, also contribute to the enhanced light output power of the LED, which indicate further
www.rsc.org/nanoscale
enhancement is manifested even in the optically ineffective sacrificial area.
Introduction GaN-based LEDs have recently garnered an increasing amount of attention in the elds of solid-state lighting, signaling, and large displays, which is expected as a complimentary replacement of traditional illumination sources based on incandescent and uorescent lights with lower energy consumption, longer lifetime, and higher efficiency light sources.1–3 Nevertheless, the reported external quantum efficiency (EQE) of GaN-based LEDs, which is intimately related to their internal quantum efficiency and light-extraction efficiency (LEE), is unsatisfactory for practical applications.4 Low LEE is a primary reason for low EQE, which is mainly attributed to the low probability of photons escaping from the interface between the LED structures and the air or resins because a large index contrast of approximately 1.5 typically results in the decrease of transmission at the interface. a
Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea. E-mail: mjeong@ skku.edu
b
Department of Energy Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea
c School of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju 561-756, Republic of Korea. E-mail: [email protected] d
Department of Physics, Jeju National University, Jeju 690-756, Republic of Korea
e
Department of Electronics and Radio Engineering, Kyung Hee University, Yongin 446701, Republic of Korea
f
School of Information and Communications, Gwangju Institute of Science and Technology, Gwangju 500-712, Republic of Korea † Electronic supplementary 10.1039/c3nr06584g
information
(ESI)
This journal is © The Royal Society of Chemistry 2014
available.
See
DOI:
Under this condition, considerable numbers of photons that are generated inside the device are reected from the interface, reabsorbed, and become internally extinct. Thus, several studies have focused on enhancing LEE by incorporating geometric structures either inside or outside the LEDs.5 For example, p-GaN and/or ITO contact layers have been intentionally roughened by etching,6 graded-refractive-index ITO,7,8 and periodic patterns, such as photonic crystal structures, have been fabricated using laser holographic,9 e-beam,10 and nanoimprint lithographies.11 However, the etching processes incited poor electrical properties,12 and lithography techniques have their own economic limitations when applied to a large area. Recently, hydrothermally grown ZnO micro-/nanostructures have been used to enhance the light output power without damaging the electrical or structural properties of the devices because the growth process of ZnO, which is based on the selfassembled process, requires no etching or lithography techniques.13–18 To explain the mechanism of such an enhancement, Kim et al.15 proposed that random scattering from the surface roughness was increased by introducing nanorod sidewalls, which correspondingly increased the photon escape probability. Ye et al.18 suggested that ZnO NRAs act as effective photon waveguides. Moreover, Zhong et al.13 employed the gradual variation of the effective refractive index model by ZnO NRAs to explain the enhancement of the LEE. These explanations are still under debates, since their experimental results are not enough to supports their models. Therefore, an accurate description of the mechanism of the enhancement of light
Nanoscale, 2014, 6, 4371–4378 | 4371
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
output power in GaN-based LEDs with ZnO nanorods and corresponding experimental conrmation would be indispensable in developing GaN-based high-brightness and high-efficiency LEDs. In previous researches, planar sapphire substrate was used for the experiment, however, patterned sapphire substrate (PSS) which was veried as highly efficient substrate for improving light output power is generally used in the commercial LED. Therefore, it is needed to study the enhancement of light output power of LED grown on PSS for practical application of industry. In this study, we used experimental and theoretical approaches to investigate the mechanism of enhancement of the light output power in GaN-based LEDs grown on PSS that were fabricated with hydrothermally grown ZnO NRAs. The robustness of our ZnO NRAs is supported by eld-emission scanning electron microscope (FE-SEM) images. The singlecrystallinity of the ZnO NRAs grown on the LED surface was conrmed by X-ray diffraction (XRD), cross-sectional transmission electron microscopy (TEM), and micro-Raman spectroscopy. The photoluminescence and transmission spectra exhibit good optical quality in the main emission wavelength of the LED. The current–voltage (I–V) and light output power as a function of the injection current were measured using a probe station system combined with a photodetector, where a significant enhancement of light output power of approximately 30% was observed without any degradation of the electrical properties. A theoretical analysis based on rigorous coupled-wave analysis (RCWA) methods and an introduction of the effective refractive index in the ZnO NRA area conrm that the gradual decrease of the index leads to the increase of the transmission.19 A novel type of confocal scanning electroluminescence microscopy (CSEM) was installed based on ber optics,17,20 and it was applied to measure the spatially resolved electroluminescence (EL) properties of the LED on the top surface and cross-section inside the device. This observation conrms that an increase of transmission at the interface between air and the LED due to the fabrication of ZnO NRAs is responsible for the enhancement of LEE. The simulation based on a two-dimensional nite-difference time-domain (FDTD) supports our CSEM results. Moreover, the EL measurement by CSEM on the bare p-type GaN layer indicates a distinct increase in the light output power, which reveals an additional contribution of ZnO NRAs in increasing the LEE at the optically ineffective area.
Experimental The LED layers were grown using MOCVD. Trimethylgallium, trimethylindium, and ammonia were used as the precursors for Ga, In, and N. H2 was used as the carrier gas, except when N was used for the InGaN MQWs and GaN barrier layers. Then, the surface of the LED epitaxial layers was partially etched down to the n-type GaN layer using inductively coupled plasma etching with Cl2/BCl3/Ar plasmas. A 200 nm-thick ITO layer was deposited as a transparent conductive layer. A Cr/Au metal pad was subsequently deposited onto the p- and n-type GaN surface to function as an electrode. The thicknesses of the Cr/Au n- and p-type electrodes were 50 and 200 nm, respectively.
4372 | Nanoscale, 2014, 6, 4371–4378
Paper
Then, the ZnO nanorods were grown over the entire surface of the fabricated LEDs using a simple noncatalytic hydrothermal method at low temperature (150 C). The experimental parameters were systematically optimized to control the morphology of the ZnO nanorods. During the hydrothermal reaction, the ZnO nanorod growth is most signicantly affected by the activity of hydrogen ions (pH) in the solution, the reaction time, and the autoclave temperature. The experimental procedure for growing the ZnO nanorods on the LED surface was designed as follows: the fabricated LED chip was cleaned using isopropyl alcohol and deionized water and then dipped into a reaction solution. To prepare the reaction solution, a measured amount of zinc acetate dihydrate [Zn(O2CCH3)2(H2O)2] was dissolved into deionized water to produce a 0.05 M solution at room temperature. NH4OH was added to the solution to produce an alkaline reaction environment (pH 9). Hydrothermal growth processes were performed for 60 min at 150 C in an autoclave. The pressure and heating rate in the autoclave were 4 atm and 3 C s1, respectively. In the CSEM system, a static current was applied to the sample using a Keithley 2400s SourceMeter during scanning. The light collected from the LED surface was delivered to a monochromator through a multimode optical ber and was detected using a thermoelectrically cooled charge-coupled device detector.
Results and discussion Fig. 1(a) presents the schematic of the complete structure of a GaN-based LED that was grown using metal-organic chemical vapor deposition (MOCVD). Our LED structure was grown on a PSS, which is typically used as a substrate in industrial applications. Using the PSS reduces the amount of dislocation density in the GaN layer and enhances the LEE in the LED chip, thus improving the EQE.21,22 A 20 nm-thick GaN nucleation layer, a 1.5 mm-thick undoped GaN layer, a 2 mm-thick n-type GaN layer, 5-pairs InGaN/GaN multiple-quantum-well (MQW) active layers, and a 200 nm-thick p-type GaN layer were subsequently grown on PSS. Using the conventional LED fabrication technique, we produced a full chip with ITO as the transparent top conductive layer and Cr/Au as the p–n-type electrode, as shown in Fig. 1(b). Aer those processes, the ZnO nanorods were grown at a low temperature over the entire surface of the fabricated GaN-based LEDs using a simple noncatalytic hydrothermal method. The nal structure of the GaN-based LED with self-assembled ZnO nanorods is illustrated in Fig. 1(c). The hydrothermal method is more benecial than other conventional techniques, such as MOCVD,23 pulsed laser deposition,24 radio-frequency magnetron sputtering,25 and anodized aluminum oxide membrane,26 in terms of the low growth temperature.27 Because of such low temperatures, this method is directly applicable to a conventionally grown GaN-based LED that is safe from thermal damage. Light scattering and optical opacity, which are caused by surface roughness or crystalline defects, introduce a substantial but undesirable increase or decrease in LEE. This change in
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Paper
Nanoscale
Fig. 2 Structural and optical properties of ZnO NRAs, (a) plane-view SEM image of ZnO NRAs on the ITO top layer; the inset shows the cross-sectional SEM image of vertically well-aligned ZnO nanorods, (b) XRD pattern of ZnO NRAs grown on the c-plane sapphire substrate, (c) LR-TEM image of a single ZnO nanorod; the top-right and bottom-left insets present an HR-TEM image and the corresponding electron diffraction pattern, respectively, (d) micro-Raman spectrum of ZnO NRAs. The inset shows the PL spectrum of ZnO NRAs measured at room temperature; a distinct peak at 380 nm corresponds to the nearband edge transition.
Fig. 1 Fabrication of the GaN-based LED with ZnO NRAs, (a) schematic of the LED structure that was grown using MOCVD, (b) threedimensional schematic of a single LED chip produced using the conventional LED fabrication technique, and (c) final structure of the GaN-based LED with hydrothermally grown ZnO NRAs.
efficiency is different from the efficiency originating from the effective refractive index modication using ZnO NRA. To avoid this substantial change in the LEE, high structural and optical quality should be guaranteed. The structural quality is examined using FE-SEM, XRD, TEM, and Raman scattering measurement. Fig. 2(a) presents an FE-SEM image of the ZnO nanorods grown on the ITO top contact layer. The inset is a magnied cross-sectional FE-SEM image of the ZnO nanorods. These images conrm that all of the ZnO nanorods were almost vertically aligned. The NRA had an average diameter of approximately 40 nm, a length of 300 nm, and a period of approximately 40 nm. Fig. 2(b) presents the XRD results for the ZnO nanorods that were grown on the c-plane sapphire substrate. The peaks observed at 34.5 and 72.8 in the spectrum corresponds to the (002) and (004) ZnO nanorod planes, respectively. The peak visible at approximately 42 is from the c-plane sapphire substrate. Interestingly, the other peaks are completely invisible, which indicates that our ZnO nanorods are homogeneously grown on the c-axis of the hexagonal crystal plane.28 The ZnO nanorod structure on the GaN-based LED top surface was investigated in further detail using TEM. Fig. 2(c) presents a low-resolution transmission election microscopy (LR-TEM) image of an individual ZnO nanorod that was washed
This journal is © The Royal Society of Chemistry 2014
off from the as-prepared product. A high-resolution TEM (HR-TEM) image and a corresponding selective area electron diffraction (SAED) pattern are shown in the top-right and bottom-le insets, respectively. These images reveal that the nanorod has a single-crystal structure with a lattice spacing d of approximately 0.51 nm along the longitudinal axis (c-axis), which is consistent with the spacing of the bulk ZnO (001) crystal planes. Those observations conrm that the nanorod has a highly crystalline single-domain wurtzite structure.29,30 The micro-Raman scattering analysis was performed to examine the crystal structure of the ZnO nanorods by exciting the ZnO NRAs with a He–Ne laser operating at a wavelength of 632.8 nm. As shown in Fig. 2(d), a peak at 439 cm1, which was assigned as the E2high optical phonon mode, was observed in the Raman spectrum, which indicates that the ZnO nanorods exhibit a hexagonal wurtzite structure.31 The optical properties of the hydrothermally grown ZnO nanorods were investigated using macro-photoluminescence (PL) spectroscopy and transmittance measurements. The inset of Fig. 2(d) presents the PL spectrum for the ZnO nanorods with an excitation by a 325 nm-wavelength He–Cd laser. A strong and sharp near-band-edge (NBE) emission is clearly observable near 382 nm. This emission is attributed to the direct recombination of free excitons. A weak emission band occurs in the wavelength range of 500–750 nm. As reported in the literature, the broad visible red–green luminescence centered at 600 nm is associated with specic defects: oxygen vacancies and the recombination of photo-generated holes with singly ionized charge states.32,33 The I–V characteristics were measured using a probe station equipped with a Keithley 2400s SourceMeter to characterize the electrical properties of the GaN-based LED with ZnO NRAs. To
Nanoscale, 2014, 6, 4371–4378 | 4373
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
measure the I–V curve, we have removed ZnO NRAs on the metal pad by simple scratching method. Fig. 3(a) presents the I–V characteristics of the LED with (lled square) and without ZnO NRAs (opened square). As shown in the gure, both plots exhibit good diode responses with a driving current of 20 mA in an operating voltage of approximately 3.8 V, which consequently indicates an acceptable electrical performance. More importantly, the two curves are almost identical to each other throughout the entire observation voltage, which is a clear signature that the electrical properties are not damaged by the ZnO NRA fabrication. This result is a distinct merit of our hydrothermal method that is operable at lower temperatures than other methods.14 Fig. 3(b) presents the light output power as a function of the injection current for the LEDs with and without ZnO NRAs. The light output power from the LED with ZnO NRAs was approximately 30% greater than that from the LED without ZnO NRAs at an operating current of 20 mA. This value is remarkable since we have used commercially optimized epilayers with PSS which readily shows higher light output power. The insets (i) and (ii) are photographs of LED emission at an injection current of 5 mA with and without ZnO NRAs, respectively. Considering that the electrical property does not change, such enhancement of the light output power is attributed to the increased LEE of the LED structure with ZnO NRAs. To explain the enhancement of the light output power, we employed the effective refractive index based on effective medium theory. Effective medium theory is a physical model
Paper
that describes the macroscopic properties of a medium, where the refractive index of the composite is the weighted average of each component scaled by its volume fraction. If the electromagnetic-radiation wavelength is considerably larger than the nanostructure size, the classical theories of inhomogeneous media presume that the material can be treated as a homogeneous substance with an effective dielectric function and refractive index. These quantities depend on the properties of the constituents and their volume fractions and sizes.34,35 In our case, the ZnO NRAs and air are considered to occupy an arbitrary volume, and each component has a certain volume fraction. The arbitrary volume that consists of ZnO NRAs and air can be regarded as an effective medium with an effective refractive index, as depicted in Fig. 4(a). Using effective medium
Effective refractive index of a complex consisting of ZnO and air. (a) Schematics of the effective medium composed of ZnO and air, (b) the refractive indices of each layer in the GaN-based LED structure with ZnO nanorods, (c) the cross-sectional SEM image of the LED with ZnO NRAs, and (d) the effective refractive index profiles calculated based on the NRA light, which was measured from the cross-sectional SEM image. (e) Transmittance spectra of ITO and ZnO NRAs grown on ITO. (f) The calculated transmittance distribution as a function of the period and wavelength of the ZnO nanorod arrays, (g) the calculated transmittance spectra for the corresponding structures, (h) the calculated angular-dependent transmittance at a wavelength of 450 nm for GaN/ITO and GaN/ITO/ZnO NRAs.
Fig. 4
Fig. 3 Electrical properties and light output power as a function of the injection current of the GaN-based LED, (a) the current vs. voltage curves and (b) the amount of light output power as a function of the injection current for GaN-based LEDs with and without ZnO nanorod arrays. The insets show emission images of (i) conventional and (ii) ZnO-nanorod-arranged LED at an injection current of 5 mA.
4374 | Nanoscale, 2014, 6, 4371–4378
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Nanoscale
theory, the effective refractive index neff of the ZnO NRAs and air is calculated as
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
neff ¼ [nZnO2fZnO + nair2(1 fZnO)]1/2
(1)
where fZnO is the volume fraction of ZnO and nZnO and nair are the refractive indices of ZnO and air, respectively.33 The refractive indices of GaN, ITO, and ZnO are approximately 2.49, 2.04, and 2.11, respectively, at a wavelength of 450 nm, which corresponds to the LED emission wavelength.36 In ZnO NRA layer, the effective refractive index was approximated to vary from 1.85–1.00, depending on the volume fraction variation caused by an inhomogeneous length of the ZnO nanorods. The resultant effective refractive indices of each layer in the LED with ZnO NRAs are illustrated in Fig. 4(b). Clearly, the refractive indices tend to decrease from the GaN layer to air because of the presence of ZnO NRAs. In a conventional LED, the index contrast between the ITO top layer and air is as large as 1.04. However, if the top layer changes to an effective medium composed of ZnO NRAs and air, the index contrast becomes negligibly small because the ZnO NRAs form a graded refractive index (GRIN) layer. The small index difference between the LED top layer and air induces an enhanced LEE because of the increased photon escape cone angle and increased transmittance at the interface. Fig. 4(c) presents the cross-sectional SEM image of ZnO NRAs that were grown on the LED with an ITO top contact layer. Because the measured deviation of the length of the ZnO nanorods is approximately 70 nm, the volume fraction of ZnO changes from 80% to 0% along the vertical direction toward the top surface of the LEDs. Fig. 4(d) presents the modeled effective refractive index as a function of the LED height according to the volume fraction variations shown in Fig. 4(c). To attribute the length deviation of ZnO NRAs, we separated ZnO NRA layer into two sub-layers. The rst layer at a height from 200 nm to 430 nm measured from the GaN/ITO interface has constant volume fraction of 80%, which corresponds to neff ¼ 1.85. In the second layer at a height from 430 nm to 500 nm, we modeled that the volume fraction of the ZnO NRAs gradually varies from 50% to 0%. This model qualitatively coincides with the cross-sectional SEM image depicted in Fig. 4(c). These gradual decreases in the refractive index cause easier photon escape from the devices in a manner that is correlated with the increased amount of transmission at the LED–air interface. It is obvious that ZnO NRAs act as GRIN materials which lead to an increase in the LEE of the LED. The transmission spectra of the bare ITO layer and those combined with ZnO NRAs are experimentally obtained to conrm such effect. These transmittance spectra are plotted in Fig. 4(e). The transmittances of the bare ITO layer and combined layer at the 450 nm, which is an emission wavelength of the GaN-based LED, were 71% and 77%, respectively. The combined layer has higher transmittance which indicates that graded refractive index introduced by ZnO NRAs positively contributes in the LEE. The transmittance was also calculated as a function of wavelength and the period of the NRAs by applying the RCWA
This journal is © The Royal Society of Chemistry 2014
method to understand our experiment more thoroughly. The period of the ZnO NRAs was varied from 40 to 100 nm, and the average height of the ZnO nanorods was 300 nm with 70 nm deviation in this calculation. The wavelength of the emitted light ranged from 350 to 650 nm, and normal incidence was assumed throughout the entire interfaces. As shown in Fig. 4(f), a clear oscillation of the transmittance as a function of the wavelength is observed, which indicates that thin-lm interference occurs in both the ITO layer and ZnO NRAs. This interference is also affected by the period of ZnO NRAs; the oscillating period increases for larger periods. In Fig. 4(g), the transmittance that was calculated at normal incidence at a period of 40 nm is represented as a function of the wavelength. The LED with ZnO NRAs exhibited a high transmittance of approximately 93% at the operating wavelengths of 440–460 nm, whereas the LED without ZnO NRAs exhibited a relatively low transmittance of approximately 85%, which shows a qualitative agreement with experiment. Fig. 4(h) presents the angle-dependent transmittance that was calculated at the wavelength of 450 nm for the LEDs with and without ZnO NRAs. The LED with ZnO NRAs exhibited a higher transmittance for both transverse electric and transverse magnetic polarizations than the LED without ZnO NRAs over a large range of angles to approximately 25 . These simulation results indicate that the origin of the enhanced light output is the increased transmittance at the LED–air interface via the ZnO NRAs. Cross-sectional CSEM was conducted to understand how this transmission enhancement affects the device performance. Because of the high depth-resolution along the surface-normal direction of confocal microscopy, the EL through the crosssectional plane can be mapped from the substrate to the top layer of the device, as shown in Fig. 5(a) and (b). Fig. 5(a) and (b) present the two-dimensional x–z-scanned CSEM images of the LEDs without and with ZnO NRAs, respectively. The white dashed lines distinguish each layer: the LED, the ZnO NRAs, and air. A remarkable feature in Fig. 5(a) is the quasi-periodic, vertically aligned interferometric EL patterns. When reecting at the PSS/LED interface, light is diffracted because the lensshaped, hexagonally arranged quasi-periodic patterns with an averaged period of d ¼ 2 mm act as an effective two-dimensional diffraction grating.37 The lights scattered from this patterned surface form an interference fringe with an averaged period of D ¼ (D/d)l, where D ¼ 3.5 mm is the distance from the PSS to the MQWs. Considering that l ¼ 450 nm, D is calculated to be 0.8 mm, which is consistent with the measured period of interference fringes. The high visibility of these fringes in the absence of ZnO NRAs is a clear signature that major populations of generated photon inside the MQWs are reected back and forth between the LED–air interface and the PSS/LED interface, and the amount of extracted photons is considerably low as a result.38 However, in the presence of ZnO NRAs on top of the LED, those interference fringes are signicantly weakened, as shown in Fig. 5(b). This result strongly indicates that the amount of photons that reect back and forth inside the LEDs is signicantly reduced because the optical properties of the air– LED interface are modied in the presence of ZnO NRAs. The
Nanoscale, 2014, 6, 4371–4378 | 4375
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
Paper
are present. These results are in good qualitative agreement with experiments, and theoretically support that ZnO NRAs lead to an increase of the LEE by means of an increase of the transmission at the interface between LED and air. The crosssectional plot of light intensity inside (marked as line I) and outside (marked as line II) are also depicted in Fig. 5(g) and (h). Similar to the experiment, the visibility of interference fringe is smaller with ZnO NRAs. Even though the simulations are not perfect matches with the experiment because crystal defects, surface morphology, and inhomogeneity of material are not taken into account in simulation, a clear increase of light intensity is observed in the cross-section at the outside (Fig. 5(h)), as amount of 28% in average. This enhancement of light intensity exactly matches with the CSEM experiment and also matches with the light output enhancement in L-I experiment (Fig. 3(b)). This result strongly supports that the transmission enhancement by the application of ZnO NRAs is the origin of the light extraction enhancement of LED. Finally, we demonstrate the additional positive contribution of ZnO NRAs in light extraction enhancement. Fig. 6(a) presents the cross-sectional schematic image of the edge of the fabricated LED chip. Most fabricated LED structures possess an
Fig. 5 Observation of the increased transmission at the interface between the LED and air. The cross-sectional x–z-scanned CSEM image of GaN-based LED (a) without and (b) with ZnO NRAs. The EL intensity distributions measured along arrows (c) I and (d) II marked in a and b, respectively. 2D-FDTD simulation images of LEDs (e) without and (f) with ZnO NRAs. The calculated light intensity distributions monitored along (g) line I and (h) line II marked in e and f, respectively.
cross-sectional plots depicted in Fig. 5(c) present those contrasts of the interference fringes more clearly. Here, the EL intensities in the presence of ZnO NRAs (red curve) and the absence of ZnO NRAs are normalized for better comparison. As shown by the black arrows in Fig. 5(a), the uctuation of the standard deviation of the EL intensity decreases by approximately 2.5 fold when there are ZnO NRAs, which again conrms that the amount of the transmission at the interface between LED and air is considerably increased. Consequently, more photons are extracted outside the LED, and the measured EL intensity also increases by approximately 27% compared to the case when no ZnO NRAs are present, as shown in the averaged EL intensity along line II (air) in Fig. 5(d). Recall that there was no signicant change in the electrical properties in the presence/absence of ZnO NRAs, and such an increase in the light intensities outside the LEDs is purely attributed to the improvement of the LEE. Fig. 5(e) and (f) show the 2-D FDTD simulations using commercial soware (FullWave from RSo). As shown in these gures, interference fringes are pronounced in LED without ZnO NRAs, whereas it gets considerably weaker when ZnO NRAs
4376 | Nanoscale, 2014, 6, 4371–4378
Fig. 6 Enhanced EL intensity at the exposed p-type GaN region. (a) The schematic of LED structure near the mesa-etched region. The CSEM images of the LEDs (b) with and (c) without ZnO NRAs at an injection current of 1 mA, (d) effective refractive index along the structure, (e) cross-sectional EL profiles of LED along line I and line II marked in b and c.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Paper
electrical safety margin area where the bare p-type GaN is exposed. This area typically occupies approximately 10% of the area of a single chip, which detracts from the overall lightemitting performance. Fig. 6(b) and (c) present the lateral CSEM image of the LED with and without ZnO NRAs at the edge of an LED chip, respectively, including a mesa-etched region. As observed in these gures, ZnO NRAs also enhance the LEE in this region. Fig. 6(d) presents the effective refractive index prole along the structure, which introduces the light extraction enhancement also in the electrical safety margin area. A clear suppression of the EL intensity at the electrical safety margin area from 2.5 to 16 mm is observed in the absence of ZnO NRAs (see black square in the Fig. 6(e)). On the contrary, in the presence of ZnO NRAs, the EL intensity is prominently increased by more than 3 times (see the red square in Fig. 6(e)). An additional remarkable feature is the intense EL emission at the edge of the mesa-etched region. This prominent emission is attributed to photons escaping from the side wall of the LEDs. This observation suggests that the enhancement of light output by ZnO NRAs occurs over the entire LED surface, including the exposed p-type GaN layer, due to the GRIN effect.
Conclusions We demonstrated the mechanism of enhanced light output in GaN-based LED grown on PSS, that has ZnO NRAs which were grown by a simple noncatalyzed maskless hydrothermal method. Using HR-TEM, micro-Raman spectroscopy, and macro-PL spectroscopy, the hydrothermally grown ZnO nanorods were veried to have high crystalline quality and good optical properties. The amount of light output power from the LED with ZnO NRAs was improved by approximately 30% compared to the conventional LED without ZnO NRAs, whereas the electrical properties of ZnO NRAs were not signicantly affected. The effective medium theory was applied to explain the role of ZnO NRAs to such enhancement. The cross-sectional CSEM images suggest that the light output is enhanced because of the increased transmission at the ITO/air interface. Such observation is supported by the calculations based on FDTD method combined with the effective medium theory. The enhancement of EL intensity in the electrical safety margin area with ZnO NRAs is also veried using CSEM. We believe that our results clearly explain the role of ZnO NRAs for light extraction enhancement on GaN-based LEDs and pave the way to highbrightness, high-efficiency GaN-based LEDs.
Acknowledgements This work was supported by the Institute for Basic Science (EM1304) in Korea.
Notes and references 1 E. F. Schubert and J. K. Kim, Science, 2005, 308, 1274–1278. 2 A. Zukauskas, M. S. Shur and R. Gaska, Introduction to SolidState Light, vol. 1, Wiley & Sons, New York, 2002.
This journal is © The Royal Society of Chemistry 2014
Nanoscale
3 K. Okamoto and Y. Kawakami, IEEE J. Sel. Top. Quantum Electron., 2009, 15, 1199–1209. 4 S. Pimputkar, J. S. Speck, S. P. DenBaars and S. Nakamura, Nat. Photonics, 2009, 3, 180–182. 5 E. Homeyer, P. Mattila, J. Oksanen, T. Sadi, H. Nykanen, S. Suihkonen, C. Symonds, J. Tulkki, F. Tuomisto, M. Sopanen and J. Bellessa, Appl. Phys. Lett., 2013, 102, 081110–081114. 6 J. H. Son, J. U. Kim, Y. H. Song, B. J. Kim, C. J. Ryu and J.-L. Lee, Adv. Mater., 2012, 24, 2259–2262. 7 J. K. Kim, S. Chhajed, M. F. Schubert, E. F. Schubert, A. J. Fischer, M. H. Crawford, J. Cho, H. Kim and C. Sone, Adv. Mater., 2008, 20, 801–804. 8 C.-Y. Fang, Y.-L. Liu, Y.-C. Lee, H.-L. Chen, D.-H. Wan and C.-C. Yu, Adv. Funct. Mater., 2013, 23, 1412–1421. 9 D.-H. Kim, C.-O. Cho, Y.-G. Roh, H. Jeon, Y. S. Park, J. Cho, J. S. Im, C. Sone, Y. Park, W. J. Choi and Q. H. Park, Appl. Phys. Lett., 2005, 87, 203508. 10 A. Chen, S. J. Chua, P. Chen, X. Y. Chen and L. K. Jian, Nanotechnology, 2006, 17, 3903–3908. 11 K.-J. Byeon, S.-Y. Hwang and H. Lee, Appl. Phys. Lett., 2007, 91, 091106. 12 X. A. Cao, S. J. Pearton, A. P. Zhang, G. T. Dang, F. Ren, R. J. Shul, L. Zhang, R. Hickman and J. M. Van Hove, Appl. Phys. Lett., 1999, 75, 2569–2571. 13 J. Zhong, H. Chen, G. Saraf, Y. Lu, C. K. Choi, J. J. Song, D. M. Mackie and H. Shen, Appl. Phys. Lett., 2007, 90, 203515. 14 S. J. An, J. H. Chae, G.-C. Yi and G. H. Park, Appl. Phys. Lett., 2008, 92, 121108. 15 K.-K. Kim, S.-d. Lee, H. Kim, J.-C. Park, S.-N. Lee, Y. Park, S.-J. Park and S.-W. Kim, Appl. Phys. Lett., 2009, 94, 071118. 16 H. Park, K.-J. Byeon, K.-Y. Yang, J.-Y. Cho and H. Lee, Nanotechnology, 2010, 21, 355304. 17 K. S. Kim, S.-M. Kim, H. Jeong, M. S. Jeong and G. Y. Jung, Adv. Funct. Mater., 2010, 20, 1076–1082. 18 B.-U. Ye, B. J. Kim, Y. H. Song, J. H. Son, H. k. Yu, M. H. Kim, J.-L. Lee and J. M. Baik, Adv. Funct. Mater., 2012, 22, 632– 639. 19 J. W. Leem, Y. Yeh and J. S. Yu, Opt. Express, 2012, 20, 4056– 4066. 20 H. Jeong, T. S. Oh, Y. S. Lee, H. G. Kim, C.-H. Hong, E.-K. Suh, O. H. Cha, H. S. Lee and M. S. Jeong, J. Phys. D: Appl. Phys., 2011, 44, 505102. 21 H. Xiao-Hui, L. Jian-Ping, F. Ya-Ying, K. Jun-Jie, H. Yang and W. Huai-Bing, IEEE Photonics Technol. Lett., 2011, 23, 944– 946. 22 Y. B. Tao, T. J. Yu, Z. Y. Yang, D. Ling, Y. Wang, Z. Z. Chen, Z. J. Yang and G. Y. Zhang, J. Cryst. Growth, 2011, 315, 183– 187. 23 J. J. Wu and D. K. P. Wong, Adv. Mater., 2007, 19, 2015–2019. 24 Y. Shen, J.-I. Hong, S. Xu, S. Lin, H. Fang, S. Zhang, Y. Ding, R. L. Snyder and Z. L. Wang, Adv. Funct. Mater., 2010, 20, 703–707. 25 Y. J. Fang, J. Sha, Z. L. Wang, Y. T. Wan, W. W. Xia and Y. W. Wang, Appl. Phys. Lett., 2011, 98, 033103. 26 H. Chik, J. Liang, S. G. Cloutier, N. Kouklin and J. M. Xu, Appl. Phys. Lett., 2004, 84, 3376–3378. Nanoscale, 2014, 6, 4371–4378 | 4377
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
27 J. H. Kim, D. Andeen and F. F. Lange, Adv. Mater., 2006, 18, 2453–2457. 28 H.-K. Fu, C.-L. Cheng, C.-H. Wang, T.-Y. Lin and Y.-F. Chen, Adv. Funct. Mater., 2009, 19, 3471–3475. 29 U. Ozgur, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov, S. Dogan, V. Avrutin, S. J. Cho and H. Morkoc, J. Appl. Phys., 2005, 98, 041301. 30 S. H. Jung, E. Oh, K. H. Lee, W. Park and S. H. Jeong, Adv. Mater., 2007, 19, 749–753. 31 S. K. Mohanta, S. Tripathy, X. H. Zhang, D. C. Kim, C. B. Soh, A. M. Yong, W. Liu and H. K. Cho, Appl. Phys. Lett., 2009, 94, 041901–041903. 32 F. K. Shan, G. X. Liu, W. J. Lee and B. C. Shin, J. Appl. Phys., 2007, 101, 053106–053108.
4378 | Nanoscale, 2014, 6, 4371–4378
Paper
33 V. A. Fonoberov, K. A. Alim, A. A. Balandin, F. Xiu and J. Liu, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 165317. 34 Y. Ono, Y. Kimura, Y. Ohta and N. Nishida, Appl. Opt., 1987, 26, 1142–1146. 35 T. C. Choy, Effective Medium Theory, Principles and Application, Oxford University Press Inc., Oxford, 1999. 36 M. Bass, C. DeCuastis, J. Enoch, V. Lakshminarayanan, G. Li, C. MacDonald, V. Mahajan and E. V. Stryland, Handbook of Optics, McGraw-Hill Professional, New York, 3rd edn, 2009. 37 S.-M. Kim, K. S. Kim, G. Y. Jung, J. H. Baek, H. Jeong and M. S. Jeong, J. Phys. D: Appl. Phys., 2009, 42, 152004. 38 H. W. Choi, C. Liu, E. Gu, G. McConnell, J. M. Girkin, I. M. Watson and M. D. Dawson, Appl. Phys. Lett., 2004, 84, 2253–2255.
This journal is © The Royal Society of Chemistry 2014
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
PAPER
View Journal | View Issue
Cite this: Nanoscale, 2014, 6, 4371
Light-extraction enhancement of a GaN-based LED covered with ZnO nanorod arrays† Hyun Jeong,a Doo Jae Park,a Hong Seok Lee,d Yeong Hwan Ko,e Jae Su Yu,e Sang-Bae Choi,f Dong-Seon Lee,f Eun-Kyung Suh*c and Mun Seok Jeong*ab We investigate the mechanism of light extraction enhancement of a GaN-based light-emitting diode (LED) grown on patterned sapphire substrate (PSS), that has ZnO nanorod arrays (NRAs) fabricated on top of the device using the hydrothermal method. We found that the light output power of the LED with ZnO NRAs increases by approximately 30% compared to the conventional LED without damaging the electrical properties of the device. We argue that the gradual decrease of the effective refractive index, which is caused by the fabrication of ZnO NRAs, is the mechanism of the observed improvement. Our argument is confirmed by cross-sectional confocal scanning electroluminescence microscopy (CSEM) and the
Received 12th December 2013 Accepted 19th January 2014
theoretical simulations, where we observed a distinct increase of the transmission at the interface between LED and air at the operation wavelength of the LED. In addition, the plane-view CSEM results indicate that ZnO NRAs, which were grown on the bare p-type GaN layer as an electrical safety margin
DOI: 10.1039/c3nr06584g
area, also contribute to the enhanced light output power of the LED, which indicate further
www.rsc.org/nanoscale
enhancement is manifested even in the optically ineffective sacrificial area.
Introduction GaN-based LEDs have recently garnered an increasing amount of attention in the elds of solid-state lighting, signaling, and large displays, which is expected as a complimentary replacement of traditional illumination sources based on incandescent and uorescent lights with lower energy consumption, longer lifetime, and higher efficiency light sources.1–3 Nevertheless, the reported external quantum efficiency (EQE) of GaN-based LEDs, which is intimately related to their internal quantum efficiency and light-extraction efficiency (LEE), is unsatisfactory for practical applications.4 Low LEE is a primary reason for low EQE, which is mainly attributed to the low probability of photons escaping from the interface between the LED structures and the air or resins because a large index contrast of approximately 1.5 typically results in the decrease of transmission at the interface. a
Center for Integrated Nanostructure Physics, Institute for Basic Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea. E-mail: mjeong@ skku.edu
b
Department of Energy Science, Sungkyunkwan University, Suwon 440-746, Republic of Korea
c School of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju 561-756, Republic of Korea. E-mail: [email protected] d
Department of Physics, Jeju National University, Jeju 690-756, Republic of Korea
e
Department of Electronics and Radio Engineering, Kyung Hee University, Yongin 446701, Republic of Korea
f
School of Information and Communications, Gwangju Institute of Science and Technology, Gwangju 500-712, Republic of Korea † Electronic supplementary 10.1039/c3nr06584g
information
(ESI)
This journal is © The Royal Society of Chemistry 2014
available.
See
DOI:
Under this condition, considerable numbers of photons that are generated inside the device are reected from the interface, reabsorbed, and become internally extinct. Thus, several studies have focused on enhancing LEE by incorporating geometric structures either inside or outside the LEDs.5 For example, p-GaN and/or ITO contact layers have been intentionally roughened by etching,6 graded-refractive-index ITO,7,8 and periodic patterns, such as photonic crystal structures, have been fabricated using laser holographic,9 e-beam,10 and nanoimprint lithographies.11 However, the etching processes incited poor electrical properties,12 and lithography techniques have their own economic limitations when applied to a large area. Recently, hydrothermally grown ZnO micro-/nanostructures have been used to enhance the light output power without damaging the electrical or structural properties of the devices because the growth process of ZnO, which is based on the selfassembled process, requires no etching or lithography techniques.13–18 To explain the mechanism of such an enhancement, Kim et al.15 proposed that random scattering from the surface roughness was increased by introducing nanorod sidewalls, which correspondingly increased the photon escape probability. Ye et al.18 suggested that ZnO NRAs act as effective photon waveguides. Moreover, Zhong et al.13 employed the gradual variation of the effective refractive index model by ZnO NRAs to explain the enhancement of the LEE. These explanations are still under debates, since their experimental results are not enough to supports their models. Therefore, an accurate description of the mechanism of the enhancement of light
Nanoscale, 2014, 6, 4371–4378 | 4371
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
output power in GaN-based LEDs with ZnO nanorods and corresponding experimental conrmation would be indispensable in developing GaN-based high-brightness and high-efficiency LEDs. In previous researches, planar sapphire substrate was used for the experiment, however, patterned sapphire substrate (PSS) which was veried as highly efficient substrate for improving light output power is generally used in the commercial LED. Therefore, it is needed to study the enhancement of light output power of LED grown on PSS for practical application of industry. In this study, we used experimental and theoretical approaches to investigate the mechanism of enhancement of the light output power in GaN-based LEDs grown on PSS that were fabricated with hydrothermally grown ZnO NRAs. The robustness of our ZnO NRAs is supported by eld-emission scanning electron microscope (FE-SEM) images. The singlecrystallinity of the ZnO NRAs grown on the LED surface was conrmed by X-ray diffraction (XRD), cross-sectional transmission electron microscopy (TEM), and micro-Raman spectroscopy. The photoluminescence and transmission spectra exhibit good optical quality in the main emission wavelength of the LED. The current–voltage (I–V) and light output power as a function of the injection current were measured using a probe station system combined with a photodetector, where a significant enhancement of light output power of approximately 30% was observed without any degradation of the electrical properties. A theoretical analysis based on rigorous coupled-wave analysis (RCWA) methods and an introduction of the effective refractive index in the ZnO NRA area conrm that the gradual decrease of the index leads to the increase of the transmission.19 A novel type of confocal scanning electroluminescence microscopy (CSEM) was installed based on ber optics,17,20 and it was applied to measure the spatially resolved electroluminescence (EL) properties of the LED on the top surface and cross-section inside the device. This observation conrms that an increase of transmission at the interface between air and the LED due to the fabrication of ZnO NRAs is responsible for the enhancement of LEE. The simulation based on a two-dimensional nite-difference time-domain (FDTD) supports our CSEM results. Moreover, the EL measurement by CSEM on the bare p-type GaN layer indicates a distinct increase in the light output power, which reveals an additional contribution of ZnO NRAs in increasing the LEE at the optically ineffective area.
Experimental The LED layers were grown using MOCVD. Trimethylgallium, trimethylindium, and ammonia were used as the precursors for Ga, In, and N. H2 was used as the carrier gas, except when N was used for the InGaN MQWs and GaN barrier layers. Then, the surface of the LED epitaxial layers was partially etched down to the n-type GaN layer using inductively coupled plasma etching with Cl2/BCl3/Ar plasmas. A 200 nm-thick ITO layer was deposited as a transparent conductive layer. A Cr/Au metal pad was subsequently deposited onto the p- and n-type GaN surface to function as an electrode. The thicknesses of the Cr/Au n- and p-type electrodes were 50 and 200 nm, respectively.
4372 | Nanoscale, 2014, 6, 4371–4378
Paper
Then, the ZnO nanorods were grown over the entire surface of the fabricated LEDs using a simple noncatalytic hydrothermal method at low temperature (150 C). The experimental parameters were systematically optimized to control the morphology of the ZnO nanorods. During the hydrothermal reaction, the ZnO nanorod growth is most signicantly affected by the activity of hydrogen ions (pH) in the solution, the reaction time, and the autoclave temperature. The experimental procedure for growing the ZnO nanorods on the LED surface was designed as follows: the fabricated LED chip was cleaned using isopropyl alcohol and deionized water and then dipped into a reaction solution. To prepare the reaction solution, a measured amount of zinc acetate dihydrate [Zn(O2CCH3)2(H2O)2] was dissolved into deionized water to produce a 0.05 M solution at room temperature. NH4OH was added to the solution to produce an alkaline reaction environment (pH 9). Hydrothermal growth processes were performed for 60 min at 150 C in an autoclave. The pressure and heating rate in the autoclave were 4 atm and 3 C s1, respectively. In the CSEM system, a static current was applied to the sample using a Keithley 2400s SourceMeter during scanning. The light collected from the LED surface was delivered to a monochromator through a multimode optical ber and was detected using a thermoelectrically cooled charge-coupled device detector.
Results and discussion Fig. 1(a) presents the schematic of the complete structure of a GaN-based LED that was grown using metal-organic chemical vapor deposition (MOCVD). Our LED structure was grown on a PSS, which is typically used as a substrate in industrial applications. Using the PSS reduces the amount of dislocation density in the GaN layer and enhances the LEE in the LED chip, thus improving the EQE.21,22 A 20 nm-thick GaN nucleation layer, a 1.5 mm-thick undoped GaN layer, a 2 mm-thick n-type GaN layer, 5-pairs InGaN/GaN multiple-quantum-well (MQW) active layers, and a 200 nm-thick p-type GaN layer were subsequently grown on PSS. Using the conventional LED fabrication technique, we produced a full chip with ITO as the transparent top conductive layer and Cr/Au as the p–n-type electrode, as shown in Fig. 1(b). Aer those processes, the ZnO nanorods were grown at a low temperature over the entire surface of the fabricated GaN-based LEDs using a simple noncatalytic hydrothermal method. The nal structure of the GaN-based LED with self-assembled ZnO nanorods is illustrated in Fig. 1(c). The hydrothermal method is more benecial than other conventional techniques, such as MOCVD,23 pulsed laser deposition,24 radio-frequency magnetron sputtering,25 and anodized aluminum oxide membrane,26 in terms of the low growth temperature.27 Because of such low temperatures, this method is directly applicable to a conventionally grown GaN-based LED that is safe from thermal damage. Light scattering and optical opacity, which are caused by surface roughness or crystalline defects, introduce a substantial but undesirable increase or decrease in LEE. This change in
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Paper
Nanoscale
Fig. 2 Structural and optical properties of ZnO NRAs, (a) plane-view SEM image of ZnO NRAs on the ITO top layer; the inset shows the cross-sectional SEM image of vertically well-aligned ZnO nanorods, (b) XRD pattern of ZnO NRAs grown on the c-plane sapphire substrate, (c) LR-TEM image of a single ZnO nanorod; the top-right and bottom-left insets present an HR-TEM image and the corresponding electron diffraction pattern, respectively, (d) micro-Raman spectrum of ZnO NRAs. The inset shows the PL spectrum of ZnO NRAs measured at room temperature; a distinct peak at 380 nm corresponds to the nearband edge transition.
Fig. 1 Fabrication of the GaN-based LED with ZnO NRAs, (a) schematic of the LED structure that was grown using MOCVD, (b) threedimensional schematic of a single LED chip produced using the conventional LED fabrication technique, and (c) final structure of the GaN-based LED with hydrothermally grown ZnO NRAs.
efficiency is different from the efficiency originating from the effective refractive index modication using ZnO NRA. To avoid this substantial change in the LEE, high structural and optical quality should be guaranteed. The structural quality is examined using FE-SEM, XRD, TEM, and Raman scattering measurement. Fig. 2(a) presents an FE-SEM image of the ZnO nanorods grown on the ITO top contact layer. The inset is a magnied cross-sectional FE-SEM image of the ZnO nanorods. These images conrm that all of the ZnO nanorods were almost vertically aligned. The NRA had an average diameter of approximately 40 nm, a length of 300 nm, and a period of approximately 40 nm. Fig. 2(b) presents the XRD results for the ZnO nanorods that were grown on the c-plane sapphire substrate. The peaks observed at 34.5 and 72.8 in the spectrum corresponds to the (002) and (004) ZnO nanorod planes, respectively. The peak visible at approximately 42 is from the c-plane sapphire substrate. Interestingly, the other peaks are completely invisible, which indicates that our ZnO nanorods are homogeneously grown on the c-axis of the hexagonal crystal plane.28 The ZnO nanorod structure on the GaN-based LED top surface was investigated in further detail using TEM. Fig. 2(c) presents a low-resolution transmission election microscopy (LR-TEM) image of an individual ZnO nanorod that was washed
This journal is © The Royal Society of Chemistry 2014
off from the as-prepared product. A high-resolution TEM (HR-TEM) image and a corresponding selective area electron diffraction (SAED) pattern are shown in the top-right and bottom-le insets, respectively. These images reveal that the nanorod has a single-crystal structure with a lattice spacing d of approximately 0.51 nm along the longitudinal axis (c-axis), which is consistent with the spacing of the bulk ZnO (001) crystal planes. Those observations conrm that the nanorod has a highly crystalline single-domain wurtzite structure.29,30 The micro-Raman scattering analysis was performed to examine the crystal structure of the ZnO nanorods by exciting the ZnO NRAs with a He–Ne laser operating at a wavelength of 632.8 nm. As shown in Fig. 2(d), a peak at 439 cm1, which was assigned as the E2high optical phonon mode, was observed in the Raman spectrum, which indicates that the ZnO nanorods exhibit a hexagonal wurtzite structure.31 The optical properties of the hydrothermally grown ZnO nanorods were investigated using macro-photoluminescence (PL) spectroscopy and transmittance measurements. The inset of Fig. 2(d) presents the PL spectrum for the ZnO nanorods with an excitation by a 325 nm-wavelength He–Cd laser. A strong and sharp near-band-edge (NBE) emission is clearly observable near 382 nm. This emission is attributed to the direct recombination of free excitons. A weak emission band occurs in the wavelength range of 500–750 nm. As reported in the literature, the broad visible red–green luminescence centered at 600 nm is associated with specic defects: oxygen vacancies and the recombination of photo-generated holes with singly ionized charge states.32,33 The I–V characteristics were measured using a probe station equipped with a Keithley 2400s SourceMeter to characterize the electrical properties of the GaN-based LED with ZnO NRAs. To
Nanoscale, 2014, 6, 4371–4378 | 4373
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
measure the I–V curve, we have removed ZnO NRAs on the metal pad by simple scratching method. Fig. 3(a) presents the I–V characteristics of the LED with (lled square) and without ZnO NRAs (opened square). As shown in the gure, both plots exhibit good diode responses with a driving current of 20 mA in an operating voltage of approximately 3.8 V, which consequently indicates an acceptable electrical performance. More importantly, the two curves are almost identical to each other throughout the entire observation voltage, which is a clear signature that the electrical properties are not damaged by the ZnO NRA fabrication. This result is a distinct merit of our hydrothermal method that is operable at lower temperatures than other methods.14 Fig. 3(b) presents the light output power as a function of the injection current for the LEDs with and without ZnO NRAs. The light output power from the LED with ZnO NRAs was approximately 30% greater than that from the LED without ZnO NRAs at an operating current of 20 mA. This value is remarkable since we have used commercially optimized epilayers with PSS which readily shows higher light output power. The insets (i) and (ii) are photographs of LED emission at an injection current of 5 mA with and without ZnO NRAs, respectively. Considering that the electrical property does not change, such enhancement of the light output power is attributed to the increased LEE of the LED structure with ZnO NRAs. To explain the enhancement of the light output power, we employed the effective refractive index based on effective medium theory. Effective medium theory is a physical model
Paper
that describes the macroscopic properties of a medium, where the refractive index of the composite is the weighted average of each component scaled by its volume fraction. If the electromagnetic-radiation wavelength is considerably larger than the nanostructure size, the classical theories of inhomogeneous media presume that the material can be treated as a homogeneous substance with an effective dielectric function and refractive index. These quantities depend on the properties of the constituents and their volume fractions and sizes.34,35 In our case, the ZnO NRAs and air are considered to occupy an arbitrary volume, and each component has a certain volume fraction. The arbitrary volume that consists of ZnO NRAs and air can be regarded as an effective medium with an effective refractive index, as depicted in Fig. 4(a). Using effective medium
Effective refractive index of a complex consisting of ZnO and air. (a) Schematics of the effective medium composed of ZnO and air, (b) the refractive indices of each layer in the GaN-based LED structure with ZnO nanorods, (c) the cross-sectional SEM image of the LED with ZnO NRAs, and (d) the effective refractive index profiles calculated based on the NRA light, which was measured from the cross-sectional SEM image. (e) Transmittance spectra of ITO and ZnO NRAs grown on ITO. (f) The calculated transmittance distribution as a function of the period and wavelength of the ZnO nanorod arrays, (g) the calculated transmittance spectra for the corresponding structures, (h) the calculated angular-dependent transmittance at a wavelength of 450 nm for GaN/ITO and GaN/ITO/ZnO NRAs.
Fig. 4
Fig. 3 Electrical properties and light output power as a function of the injection current of the GaN-based LED, (a) the current vs. voltage curves and (b) the amount of light output power as a function of the injection current for GaN-based LEDs with and without ZnO nanorod arrays. The insets show emission images of (i) conventional and (ii) ZnO-nanorod-arranged LED at an injection current of 5 mA.
4374 | Nanoscale, 2014, 6, 4371–4378
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Nanoscale
theory, the effective refractive index neff of the ZnO NRAs and air is calculated as
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
neff ¼ [nZnO2fZnO + nair2(1 fZnO)]1/2
(1)
where fZnO is the volume fraction of ZnO and nZnO and nair are the refractive indices of ZnO and air, respectively.33 The refractive indices of GaN, ITO, and ZnO are approximately 2.49, 2.04, and 2.11, respectively, at a wavelength of 450 nm, which corresponds to the LED emission wavelength.36 In ZnO NRA layer, the effective refractive index was approximated to vary from 1.85–1.00, depending on the volume fraction variation caused by an inhomogeneous length of the ZnO nanorods. The resultant effective refractive indices of each layer in the LED with ZnO NRAs are illustrated in Fig. 4(b). Clearly, the refractive indices tend to decrease from the GaN layer to air because of the presence of ZnO NRAs. In a conventional LED, the index contrast between the ITO top layer and air is as large as 1.04. However, if the top layer changes to an effective medium composed of ZnO NRAs and air, the index contrast becomes negligibly small because the ZnO NRAs form a graded refractive index (GRIN) layer. The small index difference between the LED top layer and air induces an enhanced LEE because of the increased photon escape cone angle and increased transmittance at the interface. Fig. 4(c) presents the cross-sectional SEM image of ZnO NRAs that were grown on the LED with an ITO top contact layer. Because the measured deviation of the length of the ZnO nanorods is approximately 70 nm, the volume fraction of ZnO changes from 80% to 0% along the vertical direction toward the top surface of the LEDs. Fig. 4(d) presents the modeled effective refractive index as a function of the LED height according to the volume fraction variations shown in Fig. 4(c). To attribute the length deviation of ZnO NRAs, we separated ZnO NRA layer into two sub-layers. The rst layer at a height from 200 nm to 430 nm measured from the GaN/ITO interface has constant volume fraction of 80%, which corresponds to neff ¼ 1.85. In the second layer at a height from 430 nm to 500 nm, we modeled that the volume fraction of the ZnO NRAs gradually varies from 50% to 0%. This model qualitatively coincides with the cross-sectional SEM image depicted in Fig. 4(c). These gradual decreases in the refractive index cause easier photon escape from the devices in a manner that is correlated with the increased amount of transmission at the LED–air interface. It is obvious that ZnO NRAs act as GRIN materials which lead to an increase in the LEE of the LED. The transmission spectra of the bare ITO layer and those combined with ZnO NRAs are experimentally obtained to conrm such effect. These transmittance spectra are plotted in Fig. 4(e). The transmittances of the bare ITO layer and combined layer at the 450 nm, which is an emission wavelength of the GaN-based LED, were 71% and 77%, respectively. The combined layer has higher transmittance which indicates that graded refractive index introduced by ZnO NRAs positively contributes in the LEE. The transmittance was also calculated as a function of wavelength and the period of the NRAs by applying the RCWA
This journal is © The Royal Society of Chemistry 2014
method to understand our experiment more thoroughly. The period of the ZnO NRAs was varied from 40 to 100 nm, and the average height of the ZnO nanorods was 300 nm with 70 nm deviation in this calculation. The wavelength of the emitted light ranged from 350 to 650 nm, and normal incidence was assumed throughout the entire interfaces. As shown in Fig. 4(f), a clear oscillation of the transmittance as a function of the wavelength is observed, which indicates that thin-lm interference occurs in both the ITO layer and ZnO NRAs. This interference is also affected by the period of ZnO NRAs; the oscillating period increases for larger periods. In Fig. 4(g), the transmittance that was calculated at normal incidence at a period of 40 nm is represented as a function of the wavelength. The LED with ZnO NRAs exhibited a high transmittance of approximately 93% at the operating wavelengths of 440–460 nm, whereas the LED without ZnO NRAs exhibited a relatively low transmittance of approximately 85%, which shows a qualitative agreement with experiment. Fig. 4(h) presents the angle-dependent transmittance that was calculated at the wavelength of 450 nm for the LEDs with and without ZnO NRAs. The LED with ZnO NRAs exhibited a higher transmittance for both transverse electric and transverse magnetic polarizations than the LED without ZnO NRAs over a large range of angles to approximately 25 . These simulation results indicate that the origin of the enhanced light output is the increased transmittance at the LED–air interface via the ZnO NRAs. Cross-sectional CSEM was conducted to understand how this transmission enhancement affects the device performance. Because of the high depth-resolution along the surface-normal direction of confocal microscopy, the EL through the crosssectional plane can be mapped from the substrate to the top layer of the device, as shown in Fig. 5(a) and (b). Fig. 5(a) and (b) present the two-dimensional x–z-scanned CSEM images of the LEDs without and with ZnO NRAs, respectively. The white dashed lines distinguish each layer: the LED, the ZnO NRAs, and air. A remarkable feature in Fig. 5(a) is the quasi-periodic, vertically aligned interferometric EL patterns. When reecting at the PSS/LED interface, light is diffracted because the lensshaped, hexagonally arranged quasi-periodic patterns with an averaged period of d ¼ 2 mm act as an effective two-dimensional diffraction grating.37 The lights scattered from this patterned surface form an interference fringe with an averaged period of D ¼ (D/d)l, where D ¼ 3.5 mm is the distance from the PSS to the MQWs. Considering that l ¼ 450 nm, D is calculated to be 0.8 mm, which is consistent with the measured period of interference fringes. The high visibility of these fringes in the absence of ZnO NRAs is a clear signature that major populations of generated photon inside the MQWs are reected back and forth between the LED–air interface and the PSS/LED interface, and the amount of extracted photons is considerably low as a result.38 However, in the presence of ZnO NRAs on top of the LED, those interference fringes are signicantly weakened, as shown in Fig. 5(b). This result strongly indicates that the amount of photons that reect back and forth inside the LEDs is signicantly reduced because the optical properties of the air– LED interface are modied in the presence of ZnO NRAs. The
Nanoscale, 2014, 6, 4371–4378 | 4375
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
Paper
are present. These results are in good qualitative agreement with experiments, and theoretically support that ZnO NRAs lead to an increase of the LEE by means of an increase of the transmission at the interface between LED and air. The crosssectional plot of light intensity inside (marked as line I) and outside (marked as line II) are also depicted in Fig. 5(g) and (h). Similar to the experiment, the visibility of interference fringe is smaller with ZnO NRAs. Even though the simulations are not perfect matches with the experiment because crystal defects, surface morphology, and inhomogeneity of material are not taken into account in simulation, a clear increase of light intensity is observed in the cross-section at the outside (Fig. 5(h)), as amount of 28% in average. This enhancement of light intensity exactly matches with the CSEM experiment and also matches with the light output enhancement in L-I experiment (Fig. 3(b)). This result strongly supports that the transmission enhancement by the application of ZnO NRAs is the origin of the light extraction enhancement of LED. Finally, we demonstrate the additional positive contribution of ZnO NRAs in light extraction enhancement. Fig. 6(a) presents the cross-sectional schematic image of the edge of the fabricated LED chip. Most fabricated LED structures possess an
Fig. 5 Observation of the increased transmission at the interface between the LED and air. The cross-sectional x–z-scanned CSEM image of GaN-based LED (a) without and (b) with ZnO NRAs. The EL intensity distributions measured along arrows (c) I and (d) II marked in a and b, respectively. 2D-FDTD simulation images of LEDs (e) without and (f) with ZnO NRAs. The calculated light intensity distributions monitored along (g) line I and (h) line II marked in e and f, respectively.
cross-sectional plots depicted in Fig. 5(c) present those contrasts of the interference fringes more clearly. Here, the EL intensities in the presence of ZnO NRAs (red curve) and the absence of ZnO NRAs are normalized for better comparison. As shown by the black arrows in Fig. 5(a), the uctuation of the standard deviation of the EL intensity decreases by approximately 2.5 fold when there are ZnO NRAs, which again conrms that the amount of the transmission at the interface between LED and air is considerably increased. Consequently, more photons are extracted outside the LED, and the measured EL intensity also increases by approximately 27% compared to the case when no ZnO NRAs are present, as shown in the averaged EL intensity along line II (air) in Fig. 5(d). Recall that there was no signicant change in the electrical properties in the presence/absence of ZnO NRAs, and such an increase in the light intensities outside the LEDs is purely attributed to the improvement of the LEE. Fig. 5(e) and (f) show the 2-D FDTD simulations using commercial soware (FullWave from RSo). As shown in these gures, interference fringes are pronounced in LED without ZnO NRAs, whereas it gets considerably weaker when ZnO NRAs
4376 | Nanoscale, 2014, 6, 4371–4378
Fig. 6 Enhanced EL intensity at the exposed p-type GaN region. (a) The schematic of LED structure near the mesa-etched region. The CSEM images of the LEDs (b) with and (c) without ZnO NRAs at an injection current of 1 mA, (d) effective refractive index along the structure, (e) cross-sectional EL profiles of LED along line I and line II marked in b and c.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Paper
electrical safety margin area where the bare p-type GaN is exposed. This area typically occupies approximately 10% of the area of a single chip, which detracts from the overall lightemitting performance. Fig. 6(b) and (c) present the lateral CSEM image of the LED with and without ZnO NRAs at the edge of an LED chip, respectively, including a mesa-etched region. As observed in these gures, ZnO NRAs also enhance the LEE in this region. Fig. 6(d) presents the effective refractive index prole along the structure, which introduces the light extraction enhancement also in the electrical safety margin area. A clear suppression of the EL intensity at the electrical safety margin area from 2.5 to 16 mm is observed in the absence of ZnO NRAs (see black square in the Fig. 6(e)). On the contrary, in the presence of ZnO NRAs, the EL intensity is prominently increased by more than 3 times (see the red square in Fig. 6(e)). An additional remarkable feature is the intense EL emission at the edge of the mesa-etched region. This prominent emission is attributed to photons escaping from the side wall of the LEDs. This observation suggests that the enhancement of light output by ZnO NRAs occurs over the entire LED surface, including the exposed p-type GaN layer, due to the GRIN effect.
Conclusions We demonstrated the mechanism of enhanced light output in GaN-based LED grown on PSS, that has ZnO NRAs which were grown by a simple noncatalyzed maskless hydrothermal method. Using HR-TEM, micro-Raman spectroscopy, and macro-PL spectroscopy, the hydrothermally grown ZnO nanorods were veried to have high crystalline quality and good optical properties. The amount of light output power from the LED with ZnO NRAs was improved by approximately 30% compared to the conventional LED without ZnO NRAs, whereas the electrical properties of ZnO NRAs were not signicantly affected. The effective medium theory was applied to explain the role of ZnO NRAs to such enhancement. The cross-sectional CSEM images suggest that the light output is enhanced because of the increased transmission at the ITO/air interface. Such observation is supported by the calculations based on FDTD method combined with the effective medium theory. The enhancement of EL intensity in the electrical safety margin area with ZnO NRAs is also veried using CSEM. We believe that our results clearly explain the role of ZnO NRAs for light extraction enhancement on GaN-based LEDs and pave the way to highbrightness, high-efficiency GaN-based LEDs.
Acknowledgements This work was supported by the Institute for Basic Science (EM1304) in Korea.
Notes and references 1 E. F. Schubert and J. K. Kim, Science, 2005, 308, 1274–1278. 2 A. Zukauskas, M. S. Shur and R. Gaska, Introduction to SolidState Light, vol. 1, Wiley & Sons, New York, 2002.
This journal is © The Royal Society of Chemistry 2014
Nanoscale
3 K. Okamoto and Y. Kawakami, IEEE J. Sel. Top. Quantum Electron., 2009, 15, 1199–1209. 4 S. Pimputkar, J. S. Speck, S. P. DenBaars and S. Nakamura, Nat. Photonics, 2009, 3, 180–182. 5 E. Homeyer, P. Mattila, J. Oksanen, T. Sadi, H. Nykanen, S. Suihkonen, C. Symonds, J. Tulkki, F. Tuomisto, M. Sopanen and J. Bellessa, Appl. Phys. Lett., 2013, 102, 081110–081114. 6 J. H. Son, J. U. Kim, Y. H. Song, B. J. Kim, C. J. Ryu and J.-L. Lee, Adv. Mater., 2012, 24, 2259–2262. 7 J. K. Kim, S. Chhajed, M. F. Schubert, E. F. Schubert, A. J. Fischer, M. H. Crawford, J. Cho, H. Kim and C. Sone, Adv. Mater., 2008, 20, 801–804. 8 C.-Y. Fang, Y.-L. Liu, Y.-C. Lee, H.-L. Chen, D.-H. Wan and C.-C. Yu, Adv. Funct. Mater., 2013, 23, 1412–1421. 9 D.-H. Kim, C.-O. Cho, Y.-G. Roh, H. Jeon, Y. S. Park, J. Cho, J. S. Im, C. Sone, Y. Park, W. J. Choi and Q. H. Park, Appl. Phys. Lett., 2005, 87, 203508. 10 A. Chen, S. J. Chua, P. Chen, X. Y. Chen and L. K. Jian, Nanotechnology, 2006, 17, 3903–3908. 11 K.-J. Byeon, S.-Y. Hwang and H. Lee, Appl. Phys. Lett., 2007, 91, 091106. 12 X. A. Cao, S. J. Pearton, A. P. Zhang, G. T. Dang, F. Ren, R. J. Shul, L. Zhang, R. Hickman and J. M. Van Hove, Appl. Phys. Lett., 1999, 75, 2569–2571. 13 J. Zhong, H. Chen, G. Saraf, Y. Lu, C. K. Choi, J. J. Song, D. M. Mackie and H. Shen, Appl. Phys. Lett., 2007, 90, 203515. 14 S. J. An, J. H. Chae, G.-C. Yi and G. H. Park, Appl. Phys. Lett., 2008, 92, 121108. 15 K.-K. Kim, S.-d. Lee, H. Kim, J.-C. Park, S.-N. Lee, Y. Park, S.-J. Park and S.-W. Kim, Appl. Phys. Lett., 2009, 94, 071118. 16 H. Park, K.-J. Byeon, K.-Y. Yang, J.-Y. Cho and H. Lee, Nanotechnology, 2010, 21, 355304. 17 K. S. Kim, S.-M. Kim, H. Jeong, M. S. Jeong and G. Y. Jung, Adv. Funct. Mater., 2010, 20, 1076–1082. 18 B.-U. Ye, B. J. Kim, Y. H. Song, J. H. Son, H. k. Yu, M. H. Kim, J.-L. Lee and J. M. Baik, Adv. Funct. Mater., 2012, 22, 632– 639. 19 J. W. Leem, Y. Yeh and J. S. Yu, Opt. Express, 2012, 20, 4056– 4066. 20 H. Jeong, T. S. Oh, Y. S. Lee, H. G. Kim, C.-H. Hong, E.-K. Suh, O. H. Cha, H. S. Lee and M. S. Jeong, J. Phys. D: Appl. Phys., 2011, 44, 505102. 21 H. Xiao-Hui, L. Jian-Ping, F. Ya-Ying, K. Jun-Jie, H. Yang and W. Huai-Bing, IEEE Photonics Technol. Lett., 2011, 23, 944– 946. 22 Y. B. Tao, T. J. Yu, Z. Y. Yang, D. Ling, Y. Wang, Z. Z. Chen, Z. J. Yang and G. Y. Zhang, J. Cryst. Growth, 2011, 315, 183– 187. 23 J. J. Wu and D. K. P. Wong, Adv. Mater., 2007, 19, 2015–2019. 24 Y. Shen, J.-I. Hong, S. Xu, S. Lin, H. Fang, S. Zhang, Y. Ding, R. L. Snyder and Z. L. Wang, Adv. Funct. Mater., 2010, 20, 703–707. 25 Y. J. Fang, J. Sha, Z. L. Wang, Y. T. Wan, W. W. Xia and Y. W. Wang, Appl. Phys. Lett., 2011, 98, 033103. 26 H. Chik, J. Liang, S. G. Cloutier, N. Kouklin and J. M. Xu, Appl. Phys. Lett., 2004, 84, 3376–3378. Nanoscale, 2014, 6, 4371–4378 | 4377
View Article Online
Published on 14 March 2014. Downloaded by National Dong Hwa University Library on 29/03/2014 10:43:41.
Nanoscale
27 J. H. Kim, D. Andeen and F. F. Lange, Adv. Mater., 2006, 18, 2453–2457. 28 H.-K. Fu, C.-L. Cheng, C.-H. Wang, T.-Y. Lin and Y.-F. Chen, Adv. Funct. Mater., 2009, 19, 3471–3475. 29 U. Ozgur, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov, S. Dogan, V. Avrutin, S. J. Cho and H. Morkoc, J. Appl. Phys., 2005, 98, 041301. 30 S. H. Jung, E. Oh, K. H. Lee, W. Park and S. H. Jeong, Adv. Mater., 2007, 19, 749–753. 31 S. K. Mohanta, S. Tripathy, X. H. Zhang, D. C. Kim, C. B. Soh, A. M. Yong, W. Liu and H. K. Cho, Appl. Phys. Lett., 2009, 94, 041901–041903. 32 F. K. Shan, G. X. Liu, W. J. Lee and B. C. Shin, J. Appl. Phys., 2007, 101, 053106–053108.
4378 | Nanoscale, 2014, 6, 4371–4378
Paper
33 V. A. Fonoberov, K. A. Alim, A. A. Balandin, F. Xiu and J. Liu, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 165317. 34 Y. Ono, Y. Kimura, Y. Ohta and N. Nishida, Appl. Opt., 1987, 26, 1142–1146. 35 T. C. Choy, Effective Medium Theory, Principles and Application, Oxford University Press Inc., Oxford, 1999. 36 M. Bass, C. DeCuastis, J. Enoch, V. Lakshminarayanan, G. Li, C. MacDonald, V. Mahajan and E. V. Stryland, Handbook of Optics, McGraw-Hill Professional, New York, 3rd edn, 2009. 37 S.-M. Kim, K. S. Kim, G. Y. Jung, J. H. Baek, H. Jeong and M. S. Jeong, J. Phys. D: Appl. Phys., 2009, 42, 152004. 38 H. W. Choi, C. Liu, E. Gu, G. McConnell, J. M. Girkin, I. M. Watson and M. D. Dawson, Appl. Phys. Lett., 2004, 84, 2253–2255.
This journal is © The Royal Society of Chemistry 2014
