Optical Characterization of Carrier Localization, Carrier Transportation and Carrier Recombination in Blue-Emitting InGaN/GaN MQWs

Carrier localization, transportation and recombination in blue-emitting InGaN/GaN multiple quantum wells were analyzed using temperature-dependent photoluminescence spectroscopy, confocal laser scanning microscopy and time-resolved photoluminescence (TRPL). The temperature-dependent shift of PL intensity was fitted with Arrhenius equation and explained using two non-radiative channels, which are related with thermal activation of carriers from different confining potentials. The S-shaped shift of PL peak energy and inverse-S-shaped shift of PL full width at half maximum were explained with carrier localization and carrier transportation. The TRPL spectra taken at several different places from bright region to dark region in the confocal microscopic image showed that the fast decay life-time τ1 increases with decreasing PL intensity, indicating a higher carrier transportation rate at bright region, while the slow decay life-time τ2 decreases with decreasing PL intensity, indicating a higher probability of non-radiative recombination at dark region. © The Author(s) 2014. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0011502jss] All rights reserved.

InGaN/GaN MQWs grown on c-plane sapphire substrate show high internal quantum efficiency despite the high threading dislocation (TD) density (10 8 cm −2 ∼ 10 10 cm −2 ) resulting from lattice mismatch. 1 This high-IQE phenomenon is often attributed to bandgap energy fluctuation and carrier localization in the InGaN QW layer. 2 Previous research has been focused on studying the carrier localization effect in potential minimum using transmission electron microscopy, 3 nearfield scanning optical microscopy. 4 However, not too much work has been focused on the carrier transportation between different regions in the QW layer. In the present work, temperature-dependent photoluminescence (PL), confocal laser scanning microscopy (CLSM) and time-resolved photoluminescence (TRPL) are utilized to analyze the carrier localization and transportation behavior.

Experimental
Two InGaN/GaN MQWs samples were grown on c-plane sapphire substrate in a Veeco K465i GaN metal-organic chemical vapor deposition (MOCVD) reactor. Triethylgallium (TEGa) and trimethylindium (TMIn) were used as group III sources. Ammonia (NH 3 ) was used as group V source. Nitrogen was used as carrier gas. A 30-nm-thick GaN nucleation layer was grown first on the substrate, followed by a 0.5-μm-thick GaN buffer layer and a 2-μm-thick Si-doped n-type GaN layer. This preliminary structure serves as a GaN template for further growth. Analysis of the GaN template is as follow. The X-ray diffraction (XRD) ω(002) and ω(102) rocking curves scan indicated a TD density of approximately 4.4 × 10 8 cm −2 . The atomic force microscopy (AFM) surface morphology scan of the GaN template in a 5 × 5 μm 2 area exhibited a root mean square (RMS) roughness of 0.4 nm. Four-period MQWs were then grown on the GaN template, consisting of 2.7-nm-thick InGaN quantum well layers and 12.5 nm-thick GaN barrier layers, as was confirmed by XRD. Detailed growth conditions are described elsewhere. 5 The growth temperature was also tuned to achieve a target PL peak wavelength of ∼449 nm, corresponding to an indium composition of ∼14%. Figure 1 shows the system diagram of low-temperature PL system. A Verdi-G10-semiconductor-laser pumped laser system, generating 400 nm CW laser beam, was used as excitation source. The laser beam, after transmitting through an optical fiber, was focused into a * Electrochemical Society Active Member. z E-mail: clihunc@gmail.com 2-mm-diameter spot on the sample. PL light from sample was detected by monochromator with Si detector. The sample was mounted on a cold finger in a Cryo-head, which was connected with a vacuum pump, a Helium compressor and a temperature controller, so that temperature between 10 K and 300 K can be reached. A computer was connected to the system, and a Labview program was written to control the temperature controller as well as to acquire data from monochromator and Si detector, so that the PL spectrum can be taken at any temperature between 10 K and 300 K. For nanometer scale imaging and TRPL measurement, a CLSM system, integrated with TRPL measurement system, was used. A Pico-Quant laser diode, generating 402 nm pulsed laser beam with 10 MHz frequency and 300 ps pulse width (0.3% duty cycle), was employed as excitation source. The PL emission was detected by a time-correlated single photon counting (TCSPC) avalanche photodiode (APD), which was preceded by a 445/40 nm band pass filter so that only nearbandedge emission contributes to the imaging. The microscope had a lateral spatial resolution of 200 nm and vertical spatial resolution of 100 nm. Average laser power density on the sample was 5.1 W/cm 2 , indicating an average carrier injection density of 1 × 10 17 cm −3 . This excitation condition was low enough to avoid sample heating. Details of the CLSM and TRPL system are described elsewhere. 6

Results and Discussion
Temperature-dependent PL.- Figure 2 shows the PL spectra of sample #1 measured at different temperatures. The low energy shoulder corresponds to longitudinal optical phonon replica of the main recombination peak, which is proved by the fact that it disappears at increased temperature. The high energy shoulder, on the other hand, does not appear until temperature is higher than 100 K. It corresponds to carrier recombination at weakly localized state, since carriers cannot be thermal activated to higher energy level unless the temperature is high enough.
In Figure 3 the spectrum-integrated PL intensity is plotted as a function of the temperature. It is seen that the spectrum-integrated PL intensity decreases with increasing temperature, indicating that more carriers recombine non-radiatively at higher temperature. Curve fitting done on the data in Figure 3, based on first-order standard Arrhenius equation which assumed only one non-radiative channel, did not result in good coherent to the experiment. However, fitting based on second-order Arrhenius equation , [1] which assumes two non-radiative channels with rate constant A and B and activation energy E A and E B , 7 result in good coherent with experiment data, by setting A = 15.6, B = 1.3, E A = 44.4 meV and E B = 11.3 meV, as shown in Figure 3. The contributions of each non-radiative channel are plotted separately in Figure 3. It is seen that at high temperature the non-radiative recombination is dominated by the first channel (with higher activation energy E A ), while at relatively low temperature the second non-radiative channel (with lower activation energy E B ) dominates the non-radiative recombination process. The first channel corresponds to thermal activation of carriers out of the confining potential (strongly localized state) in QW layer. A proof of this is the fact that the integrated PL intensity is significantly quenched in the temperature range where the first non-radiative channel dominates. As a result, E A actually shows the depth of the confining potential. The second non-radiative channel, on the other hand, corresponds to thermal activation of carriers out of weakly localized state. PL peak energy shifts when temperature increases from 10 K to 300 K, as shown in Figure 4a. However, different from what is predicted by Varshni's empirical expression: which states that as temperature increases semiconductor's bandgap energy shrinks, the PL peak energy in Figure 4a shows an Sshaped red-blue-redshift curve, which can be divided into three parts:  (1) below 70 K, PL peak energy decreases with increasing temperature; (2) from 70 K to 160 K, PL peak energy increases with increasing temperature; and (3) above 160 K, PL peak energy again decreases with increasing temperature. Moreover, the full width at half maximum (FWHM) of the PL spectra shows an inverse-S-shaped shift with increasing temperature, as shown in Figure 4b.
The reason for this S-shaped shift of PL peak energy and inverse-Sshaped shift of PL FWHM can be explained using carrier localization at confining potential and carrier dynamics, as shown in Figure 5.
(1) When temperature is below 70 K, the radiative recombination dominates the recombination process, so the radiative recombination life-time (τ r ) is shorter than the non-radiative recombination life-time (τ nr ). The radiative recombination life-time can be expressed as: 8 where E 2D B = 4h 2 /μ a 2D 0 2 is the quasi 2D exciton binding energy, a 2D 0 = (m 0 ε r /μ) a H is the 2D exciton Bohr radius, (T ) is the exciton linewidth at finite temperature, and all other parameters have their usual meaning. Basically, this radiative recombination life-time is temperature dependent for the (T ) and k B T terms in Equation 3. For temperature below 70 K, (T ) is much smaller than k B T .So Equation 3 can be simplified into As a result, the radiative recombination life-time is proportional to temperature. Since the carrier life-time (τ) is calculated by it is predominantly affected by the smaller life-time. Consequently, as the temperature increases from 10 K to 70 K, the radiative recombination life-time increases, leading to larger carrier life-time.
As their life-time increases, carriers can diffuse down deeper into the confining potential before recombination, as shown in Figure 5 as step (A) and (B). This reduces the recombination energy, and as a result leads to a redshift in the peak energy with increasing temperature. Moreover, as the life-time increases, carriers can diffuse to more energy states, which increases the number of available energy states during recombination, and increases the PL FWHM.
(2) When temperature is between 70 K and 160 K, non-radiative process becomes dominant during carrier recombination, and where s is carrier capture cross section, N T is the density of NRRCs, it decreases when temperature increases. Detailed calculation shows that when temperature increased from 70 K to 160 K, the nonradiative recombination life-time decreases by about 18% of the τ nr at 10 K. Consequently, carrier life-time decreases with increasing temperature at 70 K to 160 K. With increasing temperature, carriers have shorter life-time so that they cannot diffuse to the potential minimum before recombination, as shown in Figure 5 as step (C). This leads to a blueshift in the peak energy as temperature increases. On one hand, the smaller life-time reduces the number of energy states that the carriers can diffuse to; on the other hand, the higher temperature provides more thermal energy and increases the available energy states that the carriers can reach. These two factors balance each other, leading to the fact that the PL FWHM stays almost unchanged.
(3) Above 160 K, the non-radiative recombination life-time is still shorter than the radiative recombination life-time, and the nonradiative recombination still dominates the recombination process. However, in this temperature range with increasing temperature the non-radiative recombination life-time stays almost constant, as calculation show that the non-radiative recombination life-time decreases by only 6% (in contrast to case). 2 Figure 5. Explanation of red-blue-redshift of PL peak position as temperature increases.
Step (A) shows the carriers relax down to the bottom of confining potential; step (B) shows the recombination at bottom of confining potential; and step (C) shows carriers recombine before reaching the bottom of the confining potential.
As a result, the PL peak blueshift behavior caused by shorter carrier life-time becomes insignificant. On the other hand, the temperature-induced bandgap shrinkage effect, introduced by Equation 1, becomes relatively significant in this temperature range. Calculation shows that the bandgap energy shrinks by 46.1 meV when temperature increases from 160 K to 300 K, which is twice as large as the 18.3 meV bandgap shrinkage when temperature increases from 70 K to 160 K. This shrinkage in bandgap energy leads to the redshift in PL peak energy. At higher temperature, more thermal energy are provided to the carriers, so that they can hop between energy states within a much larger energy range, which increases the available number of energy states during recombination. As a result, the PL FWHM increases.
CLSM imaging and TRPL spectroscopy.- Figure 6 shows the CLSM image of the InGaN/GaN MQW sample. As analyzed before, 9 the bright regions are on the order of micrometers in size, and have smaller bandgap energy than the dark regions, due to local higher indium concentration 10 or larger quantum well thickness. 11 So carriers are localized in small bandgap energy bright regions.
TRPL spectra were taken at eight points, along the line shown in Figure 6, from bright region to dark region nearby. The TRPL spectra were fit with double exponential function and the resulting life-time data are listed in Table I. Table I Table I, it is notice that from bright region to dark region, the slow decay life-time becomes smaller. As mentioned before, the bandgap energy in bright region is smaller than that in dark region. So from point A to point H, with the decrease of the PL intensity, the bandgap energy increases, carriers are less localized and it's easier for them to escape to nonradiative recombination centers, which increases the probability/rate of non-radiative recombination. As a result, from bright region to dark region, non-radiative recombination life-time decreases and slow decay life-time decreases.
Table I also shows that the fast decay life-time increases from bright region to dark region. However, from previous analysis it is expected to see that similar as τ 2 , τ 1 also increases with increased PL intensity. Since τ 1 is related with both carrier transportation and carrier nonradiative recombination, it is believed that the carrier transportation plays an important role. In order to have increased τ 1 from bright region to dark region, the transportation rate of carriers transporting from local weakly localized state to local strongly localized states should be larger in bright region than in dark region.

Conclusions
In summary, temperature-dependent PL and microscopic TRPL were used to analyze the carrier localization and carrier transportation in blue-emitting InGaN/GaN MQWs. In temperature-dependent PL study, the change of integrated-PL intensity was fit with Arrhenius equation and explained with carrier localization in two localized states; the S-shaped PL peak energy and inverse-S-shaped PL FWHM with increasing temperature was explained with carrier localization, carrier transportation and shift in carrier life-time. In TRPL study, two lifetimes were observed in almost any place except the place with highest PL intensity. It was observed that the fast decay life-time increases with decreasing PL intensity, indicating a higher carrier transportation rate at bright region, while the slow decay life-time decreases at reduced PL intensity, indicating a higher probability of non-radiative recombination at dark region.