Effect of Nanorod Diameters on Optical Properties of GaN-Based Dual-Color Nanorod Arrays

Liang-Sen Feng(冯梁森)$^{1,2}$, Zhe Liu(刘喆)$^{1,2\hyperlink{s}{**}}$, Ning Zhang(张宁)$^{1,2}$, Bin Xue(薛斌)$^{1,2}$,\\ Jun-Xi Wang(王军喜)$^{1,2\hyperlink{s}{**}}$, Jin-Min Li(李晋闽)$^{1,2}$

doi:10.1088/0256-307X/36/2/027802

⇦Chinese Physics Letters, 2019, Vol. 36, No. 2, Article code 027802⇨ Effect of Nanorod Diameters on Optical Properties of GaN-Based Dual-Color Nanorod Arrays * Liang-Sen Feng (冯梁森)1,2, Zhe Liu (刘喆)1,2**, Ning Zhang (张宁)1,2, Bin Xue (薛斌)1,2, Jun-Xi Wang (王军喜)1,2**, Jin-Min Li (李晋闽)1,2 Affiliations 1Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083 2University of Chinese Academy of Sciences, Beijing 100049 Received 21 November 2018, online 22 January 2019 ^*Supported by the National Natural Science Foundation of China under Grant Nos 61505197 and 61334009, and the National Science and Technology Major Project under Grant No 2017YFB0403803.
^**Corresponding author. Email: jxwang@semi.ac.cn; liuzhe@semi.ac.cn Citation Text: Feng L S, Liu Z, Zhang N, Xue B and Wang J X et al 2019 Chin. Phys. Lett. 36 027802 Abstract Dual-color (blue and green) InGaN/GaN nanorod light-emitting diodes (LEDs) with three different nanorod diameters are fabricated. Enhancement of luminescence intensity per area is observed in blue and green wells, to varying degrees. When the diameter is 40 nm, it sharply decreases, which could be explained by the sidewall nonradiative recombination. Time-resolved photoluminescence is conducted to study the carrier lifetime. High recombination rate is observed in nanorod arrays, and is an order of magnitude less than that of the planar LED. When the diameter is 40 nm, the nonradiative lifetime decreases, and this explains the decrease of intensity. The 3D-FDTD simulations show the enhancement of light extraction out of geometry structure by calculating the transmittance of the nanorod arrays. DOI:10.1088/0256-307X/36/2/027802 PACS:78.55.Cr, 62.23.Hj, 85.30.-z, 78.47.jd © 2019 Chinese Physics Society Article Text GaN-based photoelectric devices, such as light-emitting diodes (LEDs), are attractive light sources for solid-state lighting. Some novel applications, such as visible lighting communications (VLCs), have become research hotspots in the past decade.[1] One of the grand challenges for these new applications is the development of phosphor-free white LEDs, including multichip white LEDs, monolithic LEDs and color-conversion white LEDs.[2] These approaches to achieve white LEDs using conventional GaN-based planar structure have been severely limited by the low efficiency and efficiency droop,[3] which can be explained by the presence of Auger recombination, defects/dislocations, polarization fields, and poor hole transport.[3,4] GaN-based nanorods are observed to possess low dislocation, low internal field and high light extraction efficiency, which are the intrinsic factors in improving photoluminescence (PL) intensity.[5,6] Moreover, the emitting units are separated and the emission surfaces face each other, so that light can be effectively mixed.[4] Thus, nanorods have been regarded as having a great potential to spectacularly improve the luminous efficiency in the green-to-red emission region and many efforts have been made in nanorod white LEDs.[3,6,7] Vertically stacking of QWs with two or three colors in nanorod array is one way to generate white light. In this study, dual-color (blue and green) nanorod arrays with three different sizes are fabricated. The nanorod structures are obtained via top-down approach, and the diameters of nanorods are 160 nm, 120 nm and 40 nm, respectively. These dichromatic light sources are very suited for VLCs because of the higher light extraction, lower carrier lifetime and more uniform field distribution. With regard to this, PL and time-resolved photoluminescence (TRPL) measures are conducted and it is found that there is a compromise between the luminescence intensity and the carrier lifetime for this application. The 3D-FDTD simulations are used to prove the enhancement of light extraction out of geometry structure. In our experiment, the samples were grown on a $c$-axis (0001) sapphire substrate through metal-organic chemical vapor deposition (MOCVD), and the MQWs consist of six periods of 2 nm InGaN/13 nm GaN blue wells (425 nm) and nine periods of 3 nm InGaN/13 nm GaN green wells (525 nm). From the aspect of crystal growth, we chose blue wells to be the buffer layers underneath green wells because the upper QW layers of blue wells are almost fully strain-relaxed[8] with larger lattice constant, which is more closed to the lattice constant of green wells. A schematic illustration of the fabrication procedures of nanorod array is shown in Fig. 1. Firstly, 200-nm-thick SiO$_{2}$ was deposited on the top of p-type GaN layer through plasma enhance vapor deposition (PECVD) as the interlayer to block Ni penetrating into p-type GaN layer. After that, 5 nm, 10 nm and 15-nm-thick Ni layers were deposited on the SiO$_{2}$ interlayer by e-beam evaporation, respectively, as illustrated in Fig. 1(b). The samples received rapid thermal annealing (RTA) treatment in N$_{2}$ at 850$^{\circ}\!$C for 60 s to obtain self-assembled Ni particles with different sizes, as shown in Fig. 1(c). Inductively coupled plasma (ICP) etching was then performed to etch SiO$_{2}$ and GaN layer below the Ni particles and the etching depth was 1 µm, which was deeply down to n-type GaN layer. After that, the nanorod structures were dipped into hot KOH solution with 5%-mass fraction for 20 min to remove side-wall damage. Finally, the Ni particles and SiO$_{2} $ interlayer were removed by buffered oxide etchant and nanorods were obtained, as illustrated in Fig. 1(d).

Fig. 1. The schematic illustration of the fabrication procedures of the dual color-nanorod LED.

Fig. 2. (a) The SEM image of Ni particles after RTA, (b) the SEM image of nanorod with the diameter of 160 nm with smooth sidewall after ICP etching, and (c) the SEM image of collapsing nanorods with smooth sidewall.

Fig. 3. PL-spectra of planar LED (black line) and nanorod LED with $d=160$ nm (red line), $d=120$ nm (green line) and $d=40$ nm (blue line) at an excitation wavelength of 325 nm at room temperature.

The thickness of Ni film is a crucial factor when fabricating nanorod structures of different sizes. Figure 2(a) shows the SEM images of Ni particles after RTA. After 60 s RTA and ICP etching procedure, nanorod arrays with diameters of 40, 120 and 160 nm were obtained. Figure 2(b) is the SEM image of nanorod. The exposed n-GaN layer at the bottom of nanorod was severely etched along $m$-plane starting from the threading dislocation, the absence of one-direction tensile stress results in tilt or collapse of some thinner nanorods on the exposed n-GaN layer as shown in Fig. 2(c). It can be observed that the nanorods have very smooth sidewalls. Blue and green lights can be reflected multiple times by these sidewalls for full mixing.[9] The micro-photoluminescence (µ-PL) measurement was conducted on planar LED and nanorod array using a 325-nm excitation laser at room temperature. The laser light source was assembled into spot with a few microns. Figure 3 shows the PL-spectra of planar LED and nanorod array at an excitation wavelength of 325 nm at room temperature, and PL luminescence of the samples with different diameters could be seen in this figure. As we can see, a blue shift occurred in both blue and green wells as the diameter decreased and the green wells showed notable blue shift of peak wavelength, indicating the strain caused by lattice mismatch between the GaN layer and InGaN layer of active region was much more released in the green wells,[10-12] excluding the quantum size effect which was only considered in smaller sizes of 15 nm.[8,12,13] A strong luminous peak can be seen at 420 nm and a relatively weak peak can be seen at 510 nm of planar LED as a result of more non-radiative recombination brought by high threading dislocation in green wells. In planar LEDs pumped by electricity, the poor hole transport along the growth direction results in weak luminescence intensity of the bottom region of the wells.[14] This will make it difficult to balance the different colors to achieve uniform white light. However, the relief of the strain in nanorod LEDs will result in smaller internal field, and more holes could transport into the wells to recombine with electrons, which would result in large enhancement of luminescence intensity and uniformity. We also calculate the µ-PL intensity per area which is more accurate when comparing the luminescence enhancement. A 2.5 µm$\times$2 µm-rectangular was designated as the light-emitting area for planar LED. The light-emitting area of nanorod array is given by $$\begin{align} A=S_{\rm side}+S_{\rm top}=\sum {(2\pi r_{i}hn_{i}+\pi r_{i}^{2}n_{i})},~~ \tag {1} \end{align} $$ where $n_{i}$ is the number of nanorods with $d=2r_{i}$, and $h$ represents the overall depth of the wells, which is 12 nm for blue wells, and 27 nm for green wells. To count accurately and briefly, we assume that there are two different diameters of nanorods for the nanorod samples. Figure 4 shows the calculated results of integrated intensity and the intensity per area for blue and green wells. For ease of comparison, the intensity per area for the planar LED was set to be 1. Overall, the PL intensity was enhanced in nanorod structures for both blue and green wells. It has been noticed that the emission per area is prominently augmented by a factor of 28 with the diameter decreased from 160 nm to 120 nm. The further stress relief in green and blue wells of MQWs, which was illustrated in many studies,[15-17] prompts more carriers to radiate visible light as the diameter decreases to 120 nm. This suggests that nanorod LEDs with smaller diameters could extract more light from quantum wells and enhance light emitting. Meanwhile, Kuo et al attributed the high efficiency to the guided mode-reduction inside a nanorod LED.[18] The number falls down to 10 and 2.4 when the diameter is 40 nm in the blue and green wells. As the diameter decreases to 40 nm, the emission area is sharply reduced and the sidewall area becomes a large proportion of emission, thus the surface state may trap carriers for nonradiative recombination.[19]

Fig. 4. The intensity ratio of integrated intensity and the intensity per area for blue (a) and green wells (b) to planar LED.

Fig. 5. The measured TRPL spectra of different diameters of nanorod LEDs with time decay.

TRPL was also conducted to explain the enhancement in PL intensity from the aspect of carrier lifetime. Figure 5 shows the TRPL spectra of the samples at room temperature with time decay. The radiative and non-radiative lifetimes can be obtained from $\tau_{\rm initial}$ and $\tau_{\rm final}$ by $$\begin{align} \tau_{\rm r}=\,&\frac{2\tau_{\rm initial}\tau_{\rm final}}{\tau_{\rm final}-\tau_{\rm initial}},~~ \tag {2} \end{align} $$ $$\begin{align} \tau_{\rm nr}=\,&2\tau_{\rm final},~~ \tag {3} \end{align} $$ where $\tau_{\rm initial}$ and $\tau_{\rm final}$ are the slopes at the initial (where ln (normalized TRPL intensity)$ < -0.3$ and final (where ln (normalized TRPL intensity)$>-3.5)$) stage.[20] Figure 6 shows the fitted results of normalized TRPL intensity and lifetimes, $\tau_{\rm initial}$ and $\tau_{\rm final}$. The radiative and non-radiative lifetimes $\tau_{\rm r}$ and $\tau_{\rm nr}$ were calculated to be 21 ns and 500 ns for planar LED, 2.5 ns and 666 ns for nanorod LEDs with $d=160$ nm, 2.2 ns and 666 ns for nanorod LEDs with $d=120$ nm, 1.6 ns and 200 ns for nanorod LEDs with $d=40$ nm. The nonradiative lifetime decreases in the $d=40$ nm sample, which means that the sidewall recombination is the reason of the intensity decrease. The data of lifetime show a trend similar to the changes of PL intensity. This shows that the reduced radiative lifetime is the main reason of PL intensity enhancement in nanorod arrays. It is necessary to mention that the nonradiative lifetime of nanorod with diameter being 40 nm is much less than others. This happens because the sidewall becomes a large proportion of emission instead of top area as the diameter decreased and the size is also far less than the diffusion length of the carriers, thus the carriers recombine promptly near the surface of the sidewall. We can see a corresponding trend of lifetime with luminescence intensity. What we need in communication applications are stronger intensity and shorter radiative lifetime. In this work, we can conclude that nanorod structure with suitable diameter can provide both conditions, and we recommend that an LED with diameter 120 nm is a good option.

Fig. 6. Fitted results of normalized TRPL intensity and lifetimes, $\tau_{\rm initial}$ and $\tau_{\rm final}$ of planar LED (a) and diameters is 160 nm (b), 120 nm (c) and 40 nm (d).

Fig. 7. The transmittance for planar LED and nanorod array with wavelength range 400–450 nm (a) and 500–550 nm (b).

To further study enhancement intensity out of the nanorod geometry, a 3D-FDTD solution is used to show the transmittance of planar LED and 120 nm-nanorod array. The simulation model is shown in Fig. 1(d), the simulation region is 4$\times$4$\times$2.5 µm$^{3}$, and a perfect matched layer (PML) boundary condition is used. One dipole source with wavelength range of 400–450 nm and 500–550 nm is set to be polarized at the center of active region in the in-plane direction. The transmittance is about 5% for planar LEDs, while for nanorod arrays are 45% and 12% for 425 nm and 525 nm, respectively, as shown in Figs. 7(a) and 7(b). In summary, dual-color nanorod LEDs with three different diameters have been fabricated by ICP etching and their luminescence characteristics have been studied. We find that the intensity of nanorod arrays is much stronger than that of planar LEDs. TRPL measurement and 3D-FDTD solutions are used to explain the reason. In TRPL measurement, high recombination rate is observed in nanorod arrays, and the carrier lifetime is an order of magnitude less than that of the planar LED. The 3D-FDTD solutions show the enhancement of light extraction out of wire geometry by calculating the transmittance of the structure. References Nitride-based semiconductors for blue and green light-emitting devicesHigh Color Rendering Index Hybrid III-Nitride/Nanocrystals White Light-Emitting DiodesControlling Electron Overflow in Phosphor-Free InGaN/GaN Nanowire White Light-Emitting DiodesWhite light emission of monolithic InGaN/GaN grown on morphology-controlled, nanostructured GaN templatesEmission color control from blue to red with nanocolumn diameter of InGaN/GaN nanocolumn arrays grown on same substrateNitrogen-polar core-shell GaN light-emitting diodes grown by selective area metalorganic vapor phase epitaxyVisible-Color-Tunable Light-Emitting DiodesThe investigation on strain relaxation and double peaks in photoluminescence of InGaN/GaN MQW layersStudy of light extraction efficiency of GaN-based light emitting diodes by using top micro/nanorod hybrid arraysOptoelectronic Properties of Nanocolumn InGaN/GaN LEDsOptical properties of InGaN/GaN nanopillars fabricated by postgrowth chemically assisted ion beam etchingStrain relaxation effect by nanotexturing InGaN/GaN multiple quantum wellStrong luminescence from strain relaxed InGaN/GaN nanotips for highly efficient light emittersA hole accelerator for InGaN/GaN light-emitting diodesStrain relaxation in GaN nanopillarsStrain relaxation in InGaN/GaN micro-pillars evidenced by high resolution cathodoluminescence hyperspectral imagingModification of far-field radiation pattern by shaping InGaN/GaN nanorodsEfficient and Directed Nano-LED Emission by a Complete Elimination of Transverse-Electric Guided ModesEffects of Strains and Defects on the Internal Quantum Efficiency of InGaN/GaN Nanorod Light Emitting DiodesAnalysis of Time-resolved Photoluminescence of InGaN Quantum Wells Using the Carrier Rate Equation

[1]	Ponce F A and Bour D P 1997 Nature 386 351
[2]	Zhuang Z, Guo X, Liu B et al 2016 Adv. Funct. Mater. 26 36
[3]	Nguyen H P T, Cui K, Zhang S et al 2012 Nano Lett. 12 1317
[4]	Song K M, Kim D H, Kim J M et al 2017 Nanotechnology 28 225703
[5]	Sekiguchi H, Kishino K and Kikuchi A 2010 Appl. Phys. Lett. 96 231104
[6]	Li S, Wang X, Fuendling S et al 2012 Appl. Phys. Lett. 101 032103
[7]	Hong Y J, Lee C H, Yoon A et al 2011 Adv. Mater. 23 3284
[8]	Zhu J H, Wang L J, Zhang S M et al 2009 J. Phys. D 42 235104
[9]	Li W, Li K, Kong F M et al 2014 Opt. Quantum Electron. 46 1413
[10]	Sacconi F, Maur M A D and Carlo A D 2012 IEEE Trans. Electron Devices 59 2979
[11]	Kawakami Y, Kaneta A, Su L et al 2010 J. Appl. Phys. 107 023522
[12]	Ramesh V, Kikuchi A, Kishino K et al 2010 J. Appl. Phys. 107 114303
[13]	Chang H J, Hsieh Y P, Chen T T et al 2007 Opt. Express 15 9357
[14]	Zhang Z H, Liu W, Tan S T et al 2014 Appl. Phys. Lett. 105 153503
[15]	Tseng W J, Gonzalez M, Dillemans L et al 2012 Appl. Phys. Lett. 101 253102
[16]	Xie E Y, Chen Z Z, Edwards P R et al 2012 J. Appl. Phys. 112 013107
[17]	Jiao Q Q, Chen Z Z, Feng Y L et al 2017 Appl. Phys. Lett. 110 052103
[18]	Kuo M L, Kim Y S, Hsieh M L et al 2011 Nano Lett. 11 476
[19]	Chang C H, Chen L Y, Huang L C et al 2012 IEEE J. Quantum Electron. 48 551
[20]	Kim H, Shin D S, Ryu H Y et al 2010 Jpn. J. Appl. Phys. 49 112402