Enabling area-selective potential-energy engineering in InGaN/GaN quantum wells by post-growth intermixing Chao Shen, Tien Khee Ng, and Boon S. Ooi* Photonics Laboratory, King Abdullah University of Science and Technology (KAUST), Thuwal, 23955-6900, Saudi Arabia * [email protected]
Abstract: We report on a unique area-selective, post-growth approach in engineering the quantum-confined potential-energy profile of InGaN/GaN quantum wells (QWs) utilizing metal/dielectric-coating induced intermixing process. This led to simultaneous realization of adjacent regions with peak emission of 2.74 eV and 2.82 eV with a high spatial resolution (~1 μm) at the coating boundary. The potential profile softening in the intermixed QW light-emitting diode (LED) was experimentally and numerically correlated, shedding light on the origin of alleviated efficiency droop from 30.5% to 16.6% (at 150 A/cm2). The technique is advantageous for fabricating high efficiency light-emitters, and is amenable to monolithic integration of nitride-based photonic devices. ©2015 Optical Society of America OCIS codes: (230.3670) Light-emitting diodes; (250.0250) Optoelectronics; (250.5590) Quantum-well, -wire and -dot devices.
References and links 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.
S. Pimputkar, J. S. Speck, S. P. DenBaars, and S. Nakamura, “Prospects for LED lighting,” Nat. Photonics 3(4), 180–182 (2009). Y. P. Zhang, Z.-H. Zhang, W. Liu, S. T. Tan, Z. G. Ju, X. L. Zhang, Y. Ji, L. C. Wang, Z. Kyaw, N. Hasanov, B. B. Zhu, S. P. Lu, X. W. Sun, and H. V. Demir, “Nonradiative recombination - critical in choosing quantum well number for InGaN/GaN light-emitting diodes,” Opt. Express 23(3), A34–A42 (2015). K. A. Denault, M. Cantore, S. Nakamura, S. P. DenBaars, and R. Seshadri, “Efficient and stable laser-driven white lighting,” AIP Adv. 3(7), 072107 (2013). S. Chen, A. Asahara, T. Ito, J. Zhang, B. Zhang, T. Suemoto, M. Yoshita, and H. Akiyama, “Gain-switching dynamics in optically pumped single-mode InGaN vertical-cavity surface-emitting lasers,” Opt. Express 22(4), 4196–4201 (2014). S. Nakamura, “Current status of GaN-based solid-state lighting,” MRS Bull. 34(2), 101–107 (2009). S. Watson, M. M. Tan, S. P. Najda, P. Perlin, M. Leszczynski, G. Targowski, S. Grzanka, and A. E. Kelly, “Visible light communications using a directly modulated 422 nm GaN laser diode,” Opt. Lett. 38(19), 3792– 3794 (2013). A. David and M. J. Grundmann, “Influence of polarization fields on carrier lifetime and recombination rates in InGaN-based light-emitting diodes,” Appl. Phys. Lett. 97(3), 033501 (2010). F. Bernardini, V. Fiorentini, and D. Vanderbilt, “Spontaneous polarization and piezoelectric constants of III-V nitrides,” Phys. Rev. B 56(16), R10024– R10027 (1997). J. Iveland, L. Martinelli, J. Peretti, J. S. Speck, and C. Weisbuch, “Direct measurement of auger electrons emitted from a semiconductor light-emitting diode under electrical injection: identification of the dominant mechanism for efficiency droop,” Phys. Rev. Lett. 110(17), 177406 (2013). F. Römer and B. Witzigmann, “Effect of Auger recombination and leakage on the droop in InGaN/GaN quantum well LEDs,” Opt. Express 22(S6 Suppl 6), A1440–A1452 (2014). Y. Zhao, S. Tanaka, C. Pan, K. Fujito, D. Feezell, J. S. Speck, S. P. DenBaars, and S. Nakamura, “High-power blue-violet semipolar (2021) InGaN/GaN light-emitting diodes with low efficiency droop at 200 A/cm2,” Appl. Phys. Express 4(8), 082104 (2011). C. Pan, S. Tanaka, F. Wu, Y. Zhao, J. S. Speck, S. Nakamura, S. P. DenBaars, and D. Feezell, “High-power, low-efficiency-droop semipolar (2021) single-quantum-well blue light-emitting diodes,” Appl. Phys. Express 5(6), 062103 (2012). H. P. Zhao and N. Tansu, “Optical gain characteristics of staggered InGaN quantum wells lasers,” J. Appl. Phys. 107(11), 113110 (2010).
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7991
14. H. Zhao, G. Liu, J. Zhang, J. D. Poplawsky, V. Dierolf, and N. Tansu, “Approaches for high internal quantum efficiency green InGaN light-emitting diodes with large overlap quantum wells,” Opt. Express 19(S4 Suppl 4), A991–A1007 (2011). 15. C. H. Wang, S. P. Chang, P. H. Ku, J. C. Li, Y. P. Lan, C. C. Lin, H. C. Yang, H. C. Kuo, T. C. Lu, S. C. Wang, and C. Y. Chang, “Hole transport improvement in InGaN/GaN light-emitting diodes by graded-composition multiple quantum barriers,” Appl. Phys. Lett. 99(17), 171106 (2011). 16. B.-C. Lin, K.-J. Chen, C.-H. Wang, C.-H. Chiu, Y.-P. Lan, C.-C. Lin, P.-T. Lee, M.-H. Shih, Y.-K. Kuo, and H.C. Kuo, “Hole injection and electron overflow improvement in InGaN/GaN light-emitting diodes by a tapered AlGaN electron blocking layer,” Opt. Express 22(1), 463–469 (2014). 17. B. S. Ooi, T. K. Ong, and O. Gunawan, “Multiple-wavelength integration in InGaAs-InGaAsP structures using pulsed laser irradiation-induced quantum-well intermixing,” IEEE J. Quantum Electron. 40(5), 481–490 (2004). 18. K. J. Zhou, Q. Jiang, Z. Y. Zhang, S. M. Chen, H. Y. Liu, Z. H. Lu, K. Kennedy, S. J. Matcher, and R. A. Hogg, “Quantum dot selective area intermixing for broadband light sources,” Opt. Express 20(24), 26950–26957 (2012). 19. J. H. G. Marsh, C. J. Hamilton, O. P. Kowalski, S. D. Mcdougall, X. Liu, and B. Qui, “Method of manufacturing optical devices and related improvements,” US Patent 20030053789 (2003). 20. C. C. Chuo, C. M. Lee, and J. I. Chyi, “Interdiffusion of In and Ga in InGaN/GaN multiple quantum wells,” Appl. Phys. Lett. 78(3), 314–316 (2001). 21. M. D. McCluskey, L. T. Romano, B. S. Krusor, N. M. Johnson, T. Suski, and J. Jun, “Interdiffusion of In and Ga in InGaN quantum wells,” Appl. Phys. Lett. 73(9), 1281–1283 (1998). 22. C. C. Chuo, C. M. Lee, T. E. Nee, and J. I. Chyi, “Effects of thermal annealing on the luminescence and structural properties of high indium-content InGaN/GaN quantum wells,” Appl. Phys. Lett. 76(26), 3902–3904 (2000). 23. H. P. Zhao, X. C. Jiao, and N. Tansu, “Analysis of Interdiffused InGaN Quantum Wells for Visible LightEmitting Diodes,” J. Disp. Technol. 9(4), 199–205 (2013). 24. K. Gelžinytė, R. Ivanov, S. Marcinkevicius, Y. Zhao, D. L. Becerra, S. Nakamura, S. P. DenBaars, and J. S. Speck, “High spatial uniformity of photoluminescence spectra in semipolar (20 2 1) plane InGaN/GaN quantum wells,” J. Appl. Phys. 117(2), 023111 (2015). 25. M. Schuster, P. O. Gervais, B. Jobst, W. Hosler, R. Averbeck, H. Riechert, A. Iberl, and R. Stommer, “Determination of the chemical composition of distorted InGaN GaN heterostructures from x-ray diffraction data,” J. Phys. D Appl. Phys. 32(10A), A56–A60 (1999). 26. H. W. Seo, L. W. Tu, Y. T. Lin, C. Y. Ho, Q. Y. Chen, L. Yuan, D. P. Norman, and N. J. Ho, “p-GaN/InGaN/nGaN pedestal nanorods: Effect of postgrowth annealing on the electrical performance,” Appl. Phys. Lett. 94(20), 201907 (2009). 27. D. G. Deppe and N. Holonyak, “Atom diffusion and impurity-induced layer disordering in quantum well III-V semiconductor heterostructures,” J. Appl. Phys. 64(12), R93–R113 (1988). 28. M. O’Neill, A. C. Bryce, J. H. Marsh, R. M. De La Rue, J. S. Roberts, and C. Jeynes, “Multiple quantum well optical waveguides with large absorption edge blue shift produced by boron and fluorine impurity-induced disordering,” Appl. Phys. Lett. 55(14), 1373–1375 (1989). 29. P. N. K. Deenapanray and C. Jagadish, “Effect of stress on impurity-free quantum well intermixing,” Electrochem. Solid-State Lett. 4(2), G11–G13 (2001). 30. J. J. Wierer, A. A. Allerman, and Q. Li, “Silicon impurity-induced layer disordering of AlGaN/AlN superlattices,” Appl. Phys. Lett. 97(5), 051907 (2010). 31. J. Crank, The Mathematics of Diffusion (Oxford University, 1979). 32. S. Birner, T. Zibold, T. Andlauer, T. Kubis, M. Sabathil, A. Trellakis, and P. Vogl, “Nextnano: general purpose 3-D simulations,” IEEE Trans. Electron. Dev. 54(9), 2137–2142 (2007). 33. B. Janjua, T. K. Ng, A. Y. Alyamani, M. M. El-Desouki, and B. S. Ooi, “Enhancement of hole confinement by monolayer insertion in asymmetric quantum-barrier UVB light emitting diodes,” IEEE Photon. J. 6(2), 1–9 (2014). 34. R. Vaxenburg, E. Lifshitz, and A. Efros, “Suppression of Auger-stimulated efficiency droop in nitride-based light emitting diodes,” Appl. Phys. Lett. 102(3), 031120 (2013).
1. Introduction The rapid development of InGaN/GaN quantum well (QW) as active region for lightemitting-diodes (LEDs) [1, 2] and laser diodes (LDs) [3, 4] has evoked its applications in solid-state lighting (SSL) [5] and visible light communication (VLC) [6]. In SSL application, it is still challenging to achieve efficient InGaN QWs due to spontaneous and piezoelectric polarization-fields induced electron-hole wavefunction overlap reduction [7, 8], and Augerstimulated efficiency droop effect [9, 10]. Recent studies suggested that the above-mentioned challenges could be partly alleviated by growing the structure lattice-matched to semi-polar bulk GaN substrate [11, 12], albeit still a costly solution. Hence, the innovations on high brightness blue-green light emitters grown on c-plane sapphire substrate are still receiving
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7992
focused attention. The engineering of potential-energy profile in active regions, using staggered [13, 14] and graded [15, 16] InGaN/GaN QWs, was proposed to mitigate the polarization fields. However, the growth of these designs requires additional efforts in monolayers compositional control (MCC), which is non-trivial since the quantum wells are typically 2-3 nm thick. In this respect, post-growth annealing presents itself as a natural and effective approach in achieving potential-profile softened InGaN/GaN QW. In VLC application, hybrid or monolithic integration of active devices (light emitters), and passive devices (detectors and modulators) are required, potentially involving multi-wavelength emitters [17] in the future. For small foot-print, low energy consumption, and high bandwidth VLC systems, as well as future edge emitting laser-based SSL applications, monolithic integration of these electro-optical (EO) devices will require emission wavelength modification in single chip, which has yet to be developed in InGaN/GaN QW system. It is noted that the compositional intermixing using thermal annealing process in InGaAs materials system has been extensively studied [18, 19]. The process promotes the interdiffusion of group-III atomic species from the QWs to adjacent quantum barriers (QBs), leading to graded compositional interface [Fig. 1(a)]. As for InGaN-based QWs, high temperature (1300-1400 °C), high N2 over pressure (up to 15 kbar) or long-period (15-40 minutes) are required for thermal-induced interdiffusion [20, 21]. However, the material dissociation and degradation, such as the formation of pyramidal defects and miscibility-gap induced In-rich clusters, associated with the above processes prevent their practical implementation, and therefore there was no report in electrically operated devices after postgrowth annealing [21, 22]. In this paper, we demonstrated the high spatial resolution, areaselective intermixing approach to realize the potential-energy engineering of InGaN/GaN QW LEDs with high material quality preserved. This was achieved by utilizing metal/dielectric capping layer in conjunction with < 1000 °C repetitive rapid thermal annealing (RRTA). 2. Experiment The InGaN/GaN QW structures used for this study were grown using metalorganic chemical vapor deposition (MOCVD) on c-plane sapphire substrates. The LED structure consists of a 3 μm Si-doped GaN layer, 12 pairs of 3 nm In0.2Ga0.8N well / 13.5 nm GaN barrier, and a 300 nm Mg-doped GaN layer. The capping layers of dielectric, and metal/dielectric were deposited using RF/DC sputtering equipment without substrate heating. For the spatial resolution study, half of the sample was coated with Molybdenum:SiO2 capping while the other half was SiO2 coated only [Fig. 1(b)]. The Mo concentration is estimated to be 4 atomicpercent. The sample was then annealed in a N2 filled RRTA chamber at 950 °C for two cycles of 120 seconds. After the intermixing process, the capping layer was removed by wet chemical etching in buffered oxide etch (BOE) solution before fabrication and measurements. The chip was then characterized using the micro-photoluminescence (PL), X-ray diffraction (XRD) and Rutherford backscattering spectroscopy (RBS) measurements. The sample was excited using a 405 nm semiconductor laser from Coherent in the LabRAM HR 800 microRaman/PL system with a spot size of < 3 µm. The excitation power on the surface of the sample was 120 mW. The estimated photo-excited carrier density is > 1019 cm−3. The chip was then fabricated into 80 µm diameter LEDs using Ni/ITO and Ti/Au as p-, and n-contact pads, respectively. The electrical, and electroluminescence (EL) characteristics of the LEDs were measured using Keithley 2400 source-measurement unit (SMU) and Ocean Optics QE 65000 spectrometer. The light output power of the LED was measured using a calibrated Si photodetector from Newport (818-UV). All the measurements were carried out at room temperature. 3. Results and discussion The PL spectra from the intermixed and non-intermixed regions are shown in Fig. 1(c). After RRTA, the emission peak energy showed a ~80 meV blueshift from ~2.74 eV in as-grown
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7993
sample to ~2.82 eV in metal/SiO2 capped region, i.e. the intermixed region. However, there is insignificant shift in SiO2 capped region (~2.74 eV), i.e. the non-intermixed region. We attribute the blueshift to the modification of the indium composition profile across the QWQB interface from a square function to an error-function [23], resulting in a softened potential profile after intermixing [Fig. 1(d)]. As expected, the FWHM (full width at half maximum) of PL peak in the non-intermixed region was found to increase from 0.160 eV (in as-grown sample) to 0.175 eV. The broad PL FWHM is mainly due to potential fluctuations related to inhomogeneous broadening arising from spatial variation in the indium composition, which can be better resolved using near-field PL measurements [24]. In our case, far-field PL measurement inadvertently collected PL signals from a larger area, and therefore the larger spatial indium compositional variation was measured. The increase in PL FWHM in the nonintermixed region indicated further increase in spatial variation of indium composition, and therefore the larger dispersion in the QW transition energy. Similar observation has been reported in annealed InGaN QWs without proper capping [18]. Interestingly, our approach led to a reduction in PL FWHM in the intermixed region (0.156 eV), suggesting that the Mo:SiO2 capping layer suppresses the increase in the inhomogeneous broadening. Concurrently, the integrated PL intensity of the intermixed region increased to ~4800 a.u. when compared to the reference as-grown sample at ~4100 a.u., indicating the possible enhancement in recombination rate in intermixed QW with reduced potential fluctuations. The integrated PL intensity for the non-intermixed region has reduced to ~2700 a.u., in conjunction with the above increase in FHWM. To further identify the spatial resolution of the intermixing process, we perform a micro-PL line-scan at 0.5 μm step-size across the two regions. We measured the fluctuation in PL peak energy in both intermixed and non-intermixed regions after RRTA, which are found to be < 10 meV. As evident in Fig. 1(d), a sharp transition of peak emission energy at the boundary with < 1 μm spatial resolution was obtained.
Fig. 1. (a) Schematic illustration of QW intermixing. (b) Schematic drawing of area-selective intermixing technique presented in this work. (c) Staggered plot of normalized PL spectra from as-grown sample (black), SiO2 capped region after RRTA (red), and Mo:SiO2 capped region after RRTA (blue). The peak blueshift was obtained in the Mo:SiO2 capped region. The peak energy values are indicated with the respective FWHM values indicated in brackets. (d) PL peak energy line-scan across the non-intermixed region (i.e. SiO2 coated), and intermixed region (i.e. Mo:SiO2 coated). The red dash lines indicate the boundary of the two regions, while the insets represent the schematic band profiles in non-intermixed region (left) and intermixed region (right).
The intermixing effect in InGaN/GaN QW as observed in micro-PL spectra is further supported using XRD and RBS measurements (Fig. 2). In the XRD θ-2θ measurements with a step-size of 0.001°, the first order InGaN peak position shifted from 33.338° in as-grown sample to 33.386° in the intermixed sample. The shift towards the GaN peak indicated a lattice constant closer to that of GaN, i.e. a decrease in the out-of-plane compressive strain
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7994
(along c-direction) due to a reduction in In mole-fraction in the QWs [25]. We observed stronger InGaN satellite peaks after RRTA and a reduction of FWHM of InGaN peak from 0.180° (as-grown) to 0.150° (intermixed) after intermixing, indicating improved crystal structures [26]. In RBS measurements, we derived the In compositions from Monte Carlo fitting, which showed the decrease from 20.8% (as-grown sample) to 18.8% after intermixing. On the other hand, there is insignificant In composition change observed in the nonintermixed region (~20.6% indium). The combined micro-PL, XRD and RBS measurements provide important evidence of In and Ga compositional interdiffusion in InGaN/GaN QWs. Our observations demonstrate the feasibility of intermixing in InGaN/GaN QW structure enabled using metal/SiO2 encapsulation, as well as the possibility of having area-selective intermixing. The intermixing mechanism is suggested to be the combination of metalimpurity enhanced interdiffusion and metal/SiO2 stress induced interdiffusion [27–30]. The latter was supported using a separate experiment based on stress measurements on the capping layers deposited on two separate 4-inch diameter silicon wafers. The tensile stress of 140 MPa was obtained on the 300 nm Mo:SiO2 capping layer (used in the intermixed region of the LED device), and the compressive stress of −300 MPa was obtained on the SiO2 capping layer with comparable thickness (as used in the non-intermixed region of the LED device). The tensile stress is expected to provide the required driving force for interdiffusion, and thus intermixing was induced with the use of a Mo:SiO2 capping layer. On the contrary, the intermixing process was suppressed in the compressive-stress SiO2 capped region.
Fig. 2. Plots of: (a) XRD θ-2θ scans, (b) RBS spectra of as-grown, non-intermixed and intermixed InGaN/GaN QWs.
We further introduce an interdiffusion model to investigate the effect of intermixing in modifying the quantum-confined transition energy, built-in strain fields and radiative recombination rates. The In concentration profile as a function of the diffusion length (Ld) along the growth direction, z, is governed by the Fick’s law [31]: C (z) =
h + z 1 h−z C0 erf + erf 2 Ld Ld
(1)
where C0 is the initial In concentration (20.8% in our study), h is half the well-width, and z = 0 is defined at the center of the well. The as-grown and intermixed band structure was modeled at 3 V forward bias condition using the effective mass method [32]. This involved solving the Poisson’s, Schrodinger, current continuity, carrier transport and photon rate equations [33]. Figure 3(a) shows the energy band line-ups of the as-grown In0.21Ga0.79N QW (black curve), and the intermixed QW with interdiffusion length of Ld = 1 Å (red curve), 5 Å (green curve) and 10 Å (blue curve). The intermixing of In and Ga atoms across the InGaN/GaN interfaces leads to the modification of the band profile, including the increasingly graded QW interface, and enlarged transition energy. It can be clearly identified that the
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7995
valence and conduction band line-ups are getting smoother in the intermixed InGaN/GaN QW with Ld = 10 Å when compared to that of the as-grown sample. The simulated transition energy shift as a function of Ld is plotted in Fig. 3(b). For our experimental measurement, the peak energy after intermixing shifted to 2.82 eV, which corresponds to Ld = 10.9 Å. The calculated electron-hole wavefunction overlap ( Γ e − hh ) in intermixed QW was found to increase from 12.7% to 15.6%. This is partially attributed to the reduction in piezoelectric field in the QWs, which is originated from the large lattice mismatch between InGaN and GaN alloy. From Fig. 3(c), the strain distribution inside the InGaN QW was modified in intermixed QW, leading to the reduction in in-plane strain tensor in intermixed structure with Ld = 10 Å. Considering the quantum efficiency of InGaN QWs, the existing large nonradiative Auger recombination (RAuger) in conventional rectangular wells reduces the EL yield, and it is considered as the one of the causes of the efficiency droop [10, 34]. Remarkably, the confining potentials in intermixed structure are smoothened, leading to a significant reduction in RAuger coefficient to 1/5 of its original value in intermixed structure at Ld = 10 Å [Fig. 3(d)]. The reduction of Auger recombination rate in potential profile softened QW was attributed to the suppression of large-momentum components of Fourier expansion in the wavefunctions of the bound carriers, as the integrals of these dominant components determines the probability of Auger recombination in III-nitride materials [34]. As a result, the efficiency droop in intermixed QW would be alleviated. In addition, the internal quantum efficiency, i.e. radiative recombination rate (Rradiative) over the total recombination rate (Rtotal), after intermixing was found to improve from 44% (as-grown, i.e. Ld = 0) to 56% at 550 A/cm2.
Fig. 3. The simulated plots of: (a) band profiles of as-grown QW and intermixed QW with diffusion length of 1 Å, 5 Å and 10 Å; (b) peak emission energy as a function of Ld (with experimental result labeled); (c) strain tensor in as-grown QW and intermixed QW with varies of diffusion lengths; (d) Auger recombination rate (RAuger) and ratio of radiative recombination rate (Rradiative) to the total recombination rate (Rtotal) as a function of Ld.
To confirm the effect of increased transition energy and reduced RAuger in intermixed QW, the electrical performance is subsequently investigated by fabricating the epi-wafer into LEDs for measuring the electroluminescence (EL) characteristics. Figure 4(a) illustrates the EL spectra of as-grown, intermixed, and non-intermixed InGaN QW at injection current of 0.4 mA. The EL curves were normalized for the ease of comparing the peak energy. The peak energy blue-shift in intermixed EL device was confirmed from the measured peak energy
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7996
from 2.73 eV to 2.80 eV. There is insignificant EL peak shift in non-intermixed LED after RRTA. The spatially (along the lateral directions) uniform alloy composition in the intermixed QW was also confirmed from the reduced EL FWHM from 0.164 eV (as-grown device) to 0.138 eV (intermixed device), while the non-intermixed QW showed the increased EL FWHM (0.171 eV), consistent with the observations in micro-PL and XRD measurements. As for the voltage-current (V-I) relation of intermixed LED (see inset in Fig. 4), there is a slight increase in turn-on voltage (from 2.9 V to 3.3 V) compared to the asgrown device. This may be partly attributed to the increased QW transition energy after intermixing. The external quantum efficiency (EQE), which is the number of photons collected from the device over the number of electrons injected, is calculated from the measured light output power and injected current, plotted in Fig. 4(b). As expected in IIInitride LEDs, the EQE reaches its peak at relative low injection current density and then rollover with increasing current injection, i.e. the efficiency droop effect, in all LEDs. Although the device after intermixing showed a slightly lower peak EQE, its efficiency exceeded that of the as-grown at the cross over injection current density value of > 31 A/cm2. A reduction in the droop characteristics, defined as (η peak − η150 A / cm2 ) / η peak , from 30.5% to 16.6% at 150 A/cm2, was observed in intermixed device when comparing with that in the as-grown case. As for the non-intermixed LED, the peak EQE dropped to 0.25 and there is no improvement in efficiency droop (34.2% at 150 A/cm2) compared to the as-grown device. To understand the above observations, we employ the ABC model, in which EQE ∝ Bn 2 / ( An + Bn 2 + Cn3 ) to analyze the carrier recombination processes after potential energy engineering. The A, B, and C denote the Shockley-Read-Hall (SRH), radiative and Auger recombination coefficients, respectively. The fitting of the measured EQE curves showed the increased A coefficient (from 8.3 × 10−6 s−1 in as-grown device to 1.0 × 10−7 s−1 in intermixed device) and decreased C coefficient (from 1.2 × 10−29 cm−3s−1 in as-grown device to 4.8 × 10−30 cm−3s−1 in intermixed device), corresponding to the SRH and Auger recombination process respectively. As the SRH recombination rate is proportional to n, while the Auger recombination rate is proportional to n3, the non-radiative recombination process is expected to be dominated by SRH recombination under low injection conditions and by Auger recombination at high injection currents. Therefore, the increased SRH recombination rate explains the lower peak EQE in potential energy engineered InGaN/GaN QW operating at relative low injection level. The possible reason for the increase is the introduction of vacancy like defects during metal/dielectric-induced interdiffusion. More importantly, the intermixed LED operating at high injection current showed a decrease in Auger recombination coefficient, which is consistent with simulation. The intermixed LED exhibited higher EQE than the as-grown device at J > 30 A/cm2. Thus the suppression of Auger related droop effect using intermixing provides a unique pathway towards higher performance LED devices.
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7997
Fig. 4. (a) EL spectra of as-grown, intermixed and non-intermixed LEDs. The peak energy values are indicated with the respective FWHM values indicated in brackets. (b) EQE vs. injection current density of as-grown, intermixed and non-intermixed InGaN/GaN QWs. Inset: voltage-current (V-I) relations of as-grown and intermixed device. The percentage of EQE droop at 150 A/cm2 is indicated.
4. Conclusion In summary, we demonstrated area-selective, and low temperature (950 °C) annealing technique based on metal/dielectric encapsulation intermixing technique for engineering the transition energy of InGaN/GaN QWs. A differential bandgap blueshift between SiO2 and metal/SiO2 encapsulated regions as large as 80 meV, corresponding to In diffusion length Ld of 10.9 Å at the QW interface, has been measured. We observed a significantly alleviated EQE droop of 14% in the intermixed QW LED, consistent with a reduction in RAuger with increasing Ld. The potential profile softened InGaN QWs after intermixing is effective in increasing the efficiency of c-plane oriented LEDs, desirable especially for high In composition structures, such as green LEDs. Our high spatial resolution, area-selective intermixing process is also applicable to monolithic-integration of photonic components in InGaN/GaN structures. Acknowledgments The authors gratefully acknowledge Yang Yang, Zaibin Guo, and Chao Zhao from Imaging and Characterization Corelab, KAUST, for assistance in PL set-up, RBS and HR-XRD measurements, respectively. The financial support King Abdulaziz City for Science and Technology (KACST) Technology Innovation Center (TIC) for Solid State Lighting at KAUST is gratefully acknowledged.
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7998
Abstract: We report on a unique area-selective, post-growth approach in engineering the quantum-confined potential-energy profile of InGaN/GaN quantum wells (QWs) utilizing metal/dielectric-coating induced intermixing process. This led to simultaneous realization of adjacent regions with peak emission of 2.74 eV and 2.82 eV with a high spatial resolution (~1 μm) at the coating boundary. The potential profile softening in the intermixed QW light-emitting diode (LED) was experimentally and numerically correlated, shedding light on the origin of alleviated efficiency droop from 30.5% to 16.6% (at 150 A/cm2). The technique is advantageous for fabricating high efficiency light-emitters, and is amenable to monolithic integration of nitride-based photonic devices. ©2015 Optical Society of America OCIS codes: (230.3670) Light-emitting diodes; (250.0250) Optoelectronics; (250.5590) Quantum-well, -wire and -dot devices.
References and links 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.
S. Pimputkar, J. S. Speck, S. P. DenBaars, and S. Nakamura, “Prospects for LED lighting,” Nat. Photonics 3(4), 180–182 (2009). Y. P. Zhang, Z.-H. Zhang, W. Liu, S. T. Tan, Z. G. Ju, X. L. Zhang, Y. Ji, L. C. Wang, Z. Kyaw, N. Hasanov, B. B. Zhu, S. P. Lu, X. W. Sun, and H. V. Demir, “Nonradiative recombination - critical in choosing quantum well number for InGaN/GaN light-emitting diodes,” Opt. Express 23(3), A34–A42 (2015). K. A. Denault, M. Cantore, S. Nakamura, S. P. DenBaars, and R. Seshadri, “Efficient and stable laser-driven white lighting,” AIP Adv. 3(7), 072107 (2013). S. Chen, A. Asahara, T. Ito, J. Zhang, B. Zhang, T. Suemoto, M. Yoshita, and H. Akiyama, “Gain-switching dynamics in optically pumped single-mode InGaN vertical-cavity surface-emitting lasers,” Opt. Express 22(4), 4196–4201 (2014). S. Nakamura, “Current status of GaN-based solid-state lighting,” MRS Bull. 34(2), 101–107 (2009). S. Watson, M. M. Tan, S. P. Najda, P. Perlin, M. Leszczynski, G. Targowski, S. Grzanka, and A. E. Kelly, “Visible light communications using a directly modulated 422 nm GaN laser diode,” Opt. Lett. 38(19), 3792– 3794 (2013). A. David and M. J. Grundmann, “Influence of polarization fields on carrier lifetime and recombination rates in InGaN-based light-emitting diodes,” Appl. Phys. Lett. 97(3), 033501 (2010). F. Bernardini, V. Fiorentini, and D. Vanderbilt, “Spontaneous polarization and piezoelectric constants of III-V nitrides,” Phys. Rev. B 56(16), R10024– R10027 (1997). J. Iveland, L. Martinelli, J. Peretti, J. S. Speck, and C. Weisbuch, “Direct measurement of auger electrons emitted from a semiconductor light-emitting diode under electrical injection: identification of the dominant mechanism for efficiency droop,” Phys. Rev. Lett. 110(17), 177406 (2013). F. Römer and B. Witzigmann, “Effect of Auger recombination and leakage on the droop in InGaN/GaN quantum well LEDs,” Opt. Express 22(S6 Suppl 6), A1440–A1452 (2014). Y. Zhao, S. Tanaka, C. Pan, K. Fujito, D. Feezell, J. S. Speck, S. P. DenBaars, and S. Nakamura, “High-power blue-violet semipolar (2021) InGaN/GaN light-emitting diodes with low efficiency droop at 200 A/cm2,” Appl. Phys. Express 4(8), 082104 (2011). C. Pan, S. Tanaka, F. Wu, Y. Zhao, J. S. Speck, S. Nakamura, S. P. DenBaars, and D. Feezell, “High-power, low-efficiency-droop semipolar (2021) single-quantum-well blue light-emitting diodes,” Appl. Phys. Express 5(6), 062103 (2012). H. P. Zhao and N. Tansu, “Optical gain characteristics of staggered InGaN quantum wells lasers,” J. Appl. Phys. 107(11), 113110 (2010).
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7991
14. H. Zhao, G. Liu, J. Zhang, J. D. Poplawsky, V. Dierolf, and N. Tansu, “Approaches for high internal quantum efficiency green InGaN light-emitting diodes with large overlap quantum wells,” Opt. Express 19(S4 Suppl 4), A991–A1007 (2011). 15. C. H. Wang, S. P. Chang, P. H. Ku, J. C. Li, Y. P. Lan, C. C. Lin, H. C. Yang, H. C. Kuo, T. C. Lu, S. C. Wang, and C. Y. Chang, “Hole transport improvement in InGaN/GaN light-emitting diodes by graded-composition multiple quantum barriers,” Appl. Phys. Lett. 99(17), 171106 (2011). 16. B.-C. Lin, K.-J. Chen, C.-H. Wang, C.-H. Chiu, Y.-P. Lan, C.-C. Lin, P.-T. Lee, M.-H. Shih, Y.-K. Kuo, and H.C. Kuo, “Hole injection and electron overflow improvement in InGaN/GaN light-emitting diodes by a tapered AlGaN electron blocking layer,” Opt. Express 22(1), 463–469 (2014). 17. B. S. Ooi, T. K. Ong, and O. Gunawan, “Multiple-wavelength integration in InGaAs-InGaAsP structures using pulsed laser irradiation-induced quantum-well intermixing,” IEEE J. Quantum Electron. 40(5), 481–490 (2004). 18. K. J. Zhou, Q. Jiang, Z. Y. Zhang, S. M. Chen, H. Y. Liu, Z. H. Lu, K. Kennedy, S. J. Matcher, and R. A. Hogg, “Quantum dot selective area intermixing for broadband light sources,” Opt. Express 20(24), 26950–26957 (2012). 19. J. H. G. Marsh, C. J. Hamilton, O. P. Kowalski, S. D. Mcdougall, X. Liu, and B. Qui, “Method of manufacturing optical devices and related improvements,” US Patent 20030053789 (2003). 20. C. C. Chuo, C. M. Lee, and J. I. Chyi, “Interdiffusion of In and Ga in InGaN/GaN multiple quantum wells,” Appl. Phys. Lett. 78(3), 314–316 (2001). 21. M. D. McCluskey, L. T. Romano, B. S. Krusor, N. M. Johnson, T. Suski, and J. Jun, “Interdiffusion of In and Ga in InGaN quantum wells,” Appl. Phys. Lett. 73(9), 1281–1283 (1998). 22. C. C. Chuo, C. M. Lee, T. E. Nee, and J. I. Chyi, “Effects of thermal annealing on the luminescence and structural properties of high indium-content InGaN/GaN quantum wells,” Appl. Phys. Lett. 76(26), 3902–3904 (2000). 23. H. P. Zhao, X. C. Jiao, and N. Tansu, “Analysis of Interdiffused InGaN Quantum Wells for Visible LightEmitting Diodes,” J. Disp. Technol. 9(4), 199–205 (2013). 24. K. Gelžinytė, R. Ivanov, S. Marcinkevicius, Y. Zhao, D. L. Becerra, S. Nakamura, S. P. DenBaars, and J. S. Speck, “High spatial uniformity of photoluminescence spectra in semipolar (20 2 1) plane InGaN/GaN quantum wells,” J. Appl. Phys. 117(2), 023111 (2015). 25. M. Schuster, P. O. Gervais, B. Jobst, W. Hosler, R. Averbeck, H. Riechert, A. Iberl, and R. Stommer, “Determination of the chemical composition of distorted InGaN GaN heterostructures from x-ray diffraction data,” J. Phys. D Appl. Phys. 32(10A), A56–A60 (1999). 26. H. W. Seo, L. W. Tu, Y. T. Lin, C. Y. Ho, Q. Y. Chen, L. Yuan, D. P. Norman, and N. J. Ho, “p-GaN/InGaN/nGaN pedestal nanorods: Effect of postgrowth annealing on the electrical performance,” Appl. Phys. Lett. 94(20), 201907 (2009). 27. D. G. Deppe and N. Holonyak, “Atom diffusion and impurity-induced layer disordering in quantum well III-V semiconductor heterostructures,” J. Appl. Phys. 64(12), R93–R113 (1988). 28. M. O’Neill, A. C. Bryce, J. H. Marsh, R. M. De La Rue, J. S. Roberts, and C. Jeynes, “Multiple quantum well optical waveguides with large absorption edge blue shift produced by boron and fluorine impurity-induced disordering,” Appl. Phys. Lett. 55(14), 1373–1375 (1989). 29. P. N. K. Deenapanray and C. Jagadish, “Effect of stress on impurity-free quantum well intermixing,” Electrochem. Solid-State Lett. 4(2), G11–G13 (2001). 30. J. J. Wierer, A. A. Allerman, and Q. Li, “Silicon impurity-induced layer disordering of AlGaN/AlN superlattices,” Appl. Phys. Lett. 97(5), 051907 (2010). 31. J. Crank, The Mathematics of Diffusion (Oxford University, 1979). 32. S. Birner, T. Zibold, T. Andlauer, T. Kubis, M. Sabathil, A. Trellakis, and P. Vogl, “Nextnano: general purpose 3-D simulations,” IEEE Trans. Electron. Dev. 54(9), 2137–2142 (2007). 33. B. Janjua, T. K. Ng, A. Y. Alyamani, M. M. El-Desouki, and B. S. Ooi, “Enhancement of hole confinement by monolayer insertion in asymmetric quantum-barrier UVB light emitting diodes,” IEEE Photon. J. 6(2), 1–9 (2014). 34. R. Vaxenburg, E. Lifshitz, and A. Efros, “Suppression of Auger-stimulated efficiency droop in nitride-based light emitting diodes,” Appl. Phys. Lett. 102(3), 031120 (2013).
1. Introduction The rapid development of InGaN/GaN quantum well (QW) as active region for lightemitting-diodes (LEDs) [1, 2] and laser diodes (LDs) [3, 4] has evoked its applications in solid-state lighting (SSL) [5] and visible light communication (VLC) [6]. In SSL application, it is still challenging to achieve efficient InGaN QWs due to spontaneous and piezoelectric polarization-fields induced electron-hole wavefunction overlap reduction [7, 8], and Augerstimulated efficiency droop effect [9, 10]. Recent studies suggested that the above-mentioned challenges could be partly alleviated by growing the structure lattice-matched to semi-polar bulk GaN substrate [11, 12], albeit still a costly solution. Hence, the innovations on high brightness blue-green light emitters grown on c-plane sapphire substrate are still receiving
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7992
focused attention. The engineering of potential-energy profile in active regions, using staggered [13, 14] and graded [15, 16] InGaN/GaN QWs, was proposed to mitigate the polarization fields. However, the growth of these designs requires additional efforts in monolayers compositional control (MCC), which is non-trivial since the quantum wells are typically 2-3 nm thick. In this respect, post-growth annealing presents itself as a natural and effective approach in achieving potential-profile softened InGaN/GaN QW. In VLC application, hybrid or monolithic integration of active devices (light emitters), and passive devices (detectors and modulators) are required, potentially involving multi-wavelength emitters [17] in the future. For small foot-print, low energy consumption, and high bandwidth VLC systems, as well as future edge emitting laser-based SSL applications, monolithic integration of these electro-optical (EO) devices will require emission wavelength modification in single chip, which has yet to be developed in InGaN/GaN QW system. It is noted that the compositional intermixing using thermal annealing process in InGaAs materials system has been extensively studied [18, 19]. The process promotes the interdiffusion of group-III atomic species from the QWs to adjacent quantum barriers (QBs), leading to graded compositional interface [Fig. 1(a)]. As for InGaN-based QWs, high temperature (1300-1400 °C), high N2 over pressure (up to 15 kbar) or long-period (15-40 minutes) are required for thermal-induced interdiffusion [20, 21]. However, the material dissociation and degradation, such as the formation of pyramidal defects and miscibility-gap induced In-rich clusters, associated with the above processes prevent their practical implementation, and therefore there was no report in electrically operated devices after postgrowth annealing [21, 22]. In this paper, we demonstrated the high spatial resolution, areaselective intermixing approach to realize the potential-energy engineering of InGaN/GaN QW LEDs with high material quality preserved. This was achieved by utilizing metal/dielectric capping layer in conjunction with < 1000 °C repetitive rapid thermal annealing (RRTA). 2. Experiment The InGaN/GaN QW structures used for this study were grown using metalorganic chemical vapor deposition (MOCVD) on c-plane sapphire substrates. The LED structure consists of a 3 μm Si-doped GaN layer, 12 pairs of 3 nm In0.2Ga0.8N well / 13.5 nm GaN barrier, and a 300 nm Mg-doped GaN layer. The capping layers of dielectric, and metal/dielectric were deposited using RF/DC sputtering equipment without substrate heating. For the spatial resolution study, half of the sample was coated with Molybdenum:SiO2 capping while the other half was SiO2 coated only [Fig. 1(b)]. The Mo concentration is estimated to be 4 atomicpercent. The sample was then annealed in a N2 filled RRTA chamber at 950 °C for two cycles of 120 seconds. After the intermixing process, the capping layer was removed by wet chemical etching in buffered oxide etch (BOE) solution before fabrication and measurements. The chip was then characterized using the micro-photoluminescence (PL), X-ray diffraction (XRD) and Rutherford backscattering spectroscopy (RBS) measurements. The sample was excited using a 405 nm semiconductor laser from Coherent in the LabRAM HR 800 microRaman/PL system with a spot size of < 3 µm. The excitation power on the surface of the sample was 120 mW. The estimated photo-excited carrier density is > 1019 cm−3. The chip was then fabricated into 80 µm diameter LEDs using Ni/ITO and Ti/Au as p-, and n-contact pads, respectively. The electrical, and electroluminescence (EL) characteristics of the LEDs were measured using Keithley 2400 source-measurement unit (SMU) and Ocean Optics QE 65000 spectrometer. The light output power of the LED was measured using a calibrated Si photodetector from Newport (818-UV). All the measurements were carried out at room temperature. 3. Results and discussion The PL spectra from the intermixed and non-intermixed regions are shown in Fig. 1(c). After RRTA, the emission peak energy showed a ~80 meV blueshift from ~2.74 eV in as-grown
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7993
sample to ~2.82 eV in metal/SiO2 capped region, i.e. the intermixed region. However, there is insignificant shift in SiO2 capped region (~2.74 eV), i.e. the non-intermixed region. We attribute the blueshift to the modification of the indium composition profile across the QWQB interface from a square function to an error-function [23], resulting in a softened potential profile after intermixing [Fig. 1(d)]. As expected, the FWHM (full width at half maximum) of PL peak in the non-intermixed region was found to increase from 0.160 eV (in as-grown sample) to 0.175 eV. The broad PL FWHM is mainly due to potential fluctuations related to inhomogeneous broadening arising from spatial variation in the indium composition, which can be better resolved using near-field PL measurements [24]. In our case, far-field PL measurement inadvertently collected PL signals from a larger area, and therefore the larger spatial indium compositional variation was measured. The increase in PL FWHM in the nonintermixed region indicated further increase in spatial variation of indium composition, and therefore the larger dispersion in the QW transition energy. Similar observation has been reported in annealed InGaN QWs without proper capping [18]. Interestingly, our approach led to a reduction in PL FWHM in the intermixed region (0.156 eV), suggesting that the Mo:SiO2 capping layer suppresses the increase in the inhomogeneous broadening. Concurrently, the integrated PL intensity of the intermixed region increased to ~4800 a.u. when compared to the reference as-grown sample at ~4100 a.u., indicating the possible enhancement in recombination rate in intermixed QW with reduced potential fluctuations. The integrated PL intensity for the non-intermixed region has reduced to ~2700 a.u., in conjunction with the above increase in FHWM. To further identify the spatial resolution of the intermixing process, we perform a micro-PL line-scan at 0.5 μm step-size across the two regions. We measured the fluctuation in PL peak energy in both intermixed and non-intermixed regions after RRTA, which are found to be < 10 meV. As evident in Fig. 1(d), a sharp transition of peak emission energy at the boundary with < 1 μm spatial resolution was obtained.
Fig. 1. (a) Schematic illustration of QW intermixing. (b) Schematic drawing of area-selective intermixing technique presented in this work. (c) Staggered plot of normalized PL spectra from as-grown sample (black), SiO2 capped region after RRTA (red), and Mo:SiO2 capped region after RRTA (blue). The peak blueshift was obtained in the Mo:SiO2 capped region. The peak energy values are indicated with the respective FWHM values indicated in brackets. (d) PL peak energy line-scan across the non-intermixed region (i.e. SiO2 coated), and intermixed region (i.e. Mo:SiO2 coated). The red dash lines indicate the boundary of the two regions, while the insets represent the schematic band profiles in non-intermixed region (left) and intermixed region (right).
The intermixing effect in InGaN/GaN QW as observed in micro-PL spectra is further supported using XRD and RBS measurements (Fig. 2). In the XRD θ-2θ measurements with a step-size of 0.001°, the first order InGaN peak position shifted from 33.338° in as-grown sample to 33.386° in the intermixed sample. The shift towards the GaN peak indicated a lattice constant closer to that of GaN, i.e. a decrease in the out-of-plane compressive strain
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7994
(along c-direction) due to a reduction in In mole-fraction in the QWs [25]. We observed stronger InGaN satellite peaks after RRTA and a reduction of FWHM of InGaN peak from 0.180° (as-grown) to 0.150° (intermixed) after intermixing, indicating improved crystal structures [26]. In RBS measurements, we derived the In compositions from Monte Carlo fitting, which showed the decrease from 20.8% (as-grown sample) to 18.8% after intermixing. On the other hand, there is insignificant In composition change observed in the nonintermixed region (~20.6% indium). The combined micro-PL, XRD and RBS measurements provide important evidence of In and Ga compositional interdiffusion in InGaN/GaN QWs. Our observations demonstrate the feasibility of intermixing in InGaN/GaN QW structure enabled using metal/SiO2 encapsulation, as well as the possibility of having area-selective intermixing. The intermixing mechanism is suggested to be the combination of metalimpurity enhanced interdiffusion and metal/SiO2 stress induced interdiffusion [27–30]. The latter was supported using a separate experiment based on stress measurements on the capping layers deposited on two separate 4-inch diameter silicon wafers. The tensile stress of 140 MPa was obtained on the 300 nm Mo:SiO2 capping layer (used in the intermixed region of the LED device), and the compressive stress of −300 MPa was obtained on the SiO2 capping layer with comparable thickness (as used in the non-intermixed region of the LED device). The tensile stress is expected to provide the required driving force for interdiffusion, and thus intermixing was induced with the use of a Mo:SiO2 capping layer. On the contrary, the intermixing process was suppressed in the compressive-stress SiO2 capped region.
Fig. 2. Plots of: (a) XRD θ-2θ scans, (b) RBS spectra of as-grown, non-intermixed and intermixed InGaN/GaN QWs.
We further introduce an interdiffusion model to investigate the effect of intermixing in modifying the quantum-confined transition energy, built-in strain fields and radiative recombination rates. The In concentration profile as a function of the diffusion length (Ld) along the growth direction, z, is governed by the Fick’s law [31]: C (z) =
h + z 1 h−z C0 erf + erf 2 Ld Ld
(1)
where C0 is the initial In concentration (20.8% in our study), h is half the well-width, and z = 0 is defined at the center of the well. The as-grown and intermixed band structure was modeled at 3 V forward bias condition using the effective mass method [32]. This involved solving the Poisson’s, Schrodinger, current continuity, carrier transport and photon rate equations [33]. Figure 3(a) shows the energy band line-ups of the as-grown In0.21Ga0.79N QW (black curve), and the intermixed QW with interdiffusion length of Ld = 1 Å (red curve), 5 Å (green curve) and 10 Å (blue curve). The intermixing of In and Ga atoms across the InGaN/GaN interfaces leads to the modification of the band profile, including the increasingly graded QW interface, and enlarged transition energy. It can be clearly identified that the
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7995
valence and conduction band line-ups are getting smoother in the intermixed InGaN/GaN QW with Ld = 10 Å when compared to that of the as-grown sample. The simulated transition energy shift as a function of Ld is plotted in Fig. 3(b). For our experimental measurement, the peak energy after intermixing shifted to 2.82 eV, which corresponds to Ld = 10.9 Å. The calculated electron-hole wavefunction overlap ( Γ e − hh ) in intermixed QW was found to increase from 12.7% to 15.6%. This is partially attributed to the reduction in piezoelectric field in the QWs, which is originated from the large lattice mismatch between InGaN and GaN alloy. From Fig. 3(c), the strain distribution inside the InGaN QW was modified in intermixed QW, leading to the reduction in in-plane strain tensor in intermixed structure with Ld = 10 Å. Considering the quantum efficiency of InGaN QWs, the existing large nonradiative Auger recombination (RAuger) in conventional rectangular wells reduces the EL yield, and it is considered as the one of the causes of the efficiency droop [10, 34]. Remarkably, the confining potentials in intermixed structure are smoothened, leading to a significant reduction in RAuger coefficient to 1/5 of its original value in intermixed structure at Ld = 10 Å [Fig. 3(d)]. The reduction of Auger recombination rate in potential profile softened QW was attributed to the suppression of large-momentum components of Fourier expansion in the wavefunctions of the bound carriers, as the integrals of these dominant components determines the probability of Auger recombination in III-nitride materials [34]. As a result, the efficiency droop in intermixed QW would be alleviated. In addition, the internal quantum efficiency, i.e. radiative recombination rate (Rradiative) over the total recombination rate (Rtotal), after intermixing was found to improve from 44% (as-grown, i.e. Ld = 0) to 56% at 550 A/cm2.
Fig. 3. The simulated plots of: (a) band profiles of as-grown QW and intermixed QW with diffusion length of 1 Å, 5 Å and 10 Å; (b) peak emission energy as a function of Ld (with experimental result labeled); (c) strain tensor in as-grown QW and intermixed QW with varies of diffusion lengths; (d) Auger recombination rate (RAuger) and ratio of radiative recombination rate (Rradiative) to the total recombination rate (Rtotal) as a function of Ld.
To confirm the effect of increased transition energy and reduced RAuger in intermixed QW, the electrical performance is subsequently investigated by fabricating the epi-wafer into LEDs for measuring the electroluminescence (EL) characteristics. Figure 4(a) illustrates the EL spectra of as-grown, intermixed, and non-intermixed InGaN QW at injection current of 0.4 mA. The EL curves were normalized for the ease of comparing the peak energy. The peak energy blue-shift in intermixed EL device was confirmed from the measured peak energy
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7996
from 2.73 eV to 2.80 eV. There is insignificant EL peak shift in non-intermixed LED after RRTA. The spatially (along the lateral directions) uniform alloy composition in the intermixed QW was also confirmed from the reduced EL FWHM from 0.164 eV (as-grown device) to 0.138 eV (intermixed device), while the non-intermixed QW showed the increased EL FWHM (0.171 eV), consistent with the observations in micro-PL and XRD measurements. As for the voltage-current (V-I) relation of intermixed LED (see inset in Fig. 4), there is a slight increase in turn-on voltage (from 2.9 V to 3.3 V) compared to the asgrown device. This may be partly attributed to the increased QW transition energy after intermixing. The external quantum efficiency (EQE), which is the number of photons collected from the device over the number of electrons injected, is calculated from the measured light output power and injected current, plotted in Fig. 4(b). As expected in IIInitride LEDs, the EQE reaches its peak at relative low injection current density and then rollover with increasing current injection, i.e. the efficiency droop effect, in all LEDs. Although the device after intermixing showed a slightly lower peak EQE, its efficiency exceeded that of the as-grown at the cross over injection current density value of > 31 A/cm2. A reduction in the droop characteristics, defined as (η peak − η150 A / cm2 ) / η peak , from 30.5% to 16.6% at 150 A/cm2, was observed in intermixed device when comparing with that in the as-grown case. As for the non-intermixed LED, the peak EQE dropped to 0.25 and there is no improvement in efficiency droop (34.2% at 150 A/cm2) compared to the as-grown device. To understand the above observations, we employ the ABC model, in which EQE ∝ Bn 2 / ( An + Bn 2 + Cn3 ) to analyze the carrier recombination processes after potential energy engineering. The A, B, and C denote the Shockley-Read-Hall (SRH), radiative and Auger recombination coefficients, respectively. The fitting of the measured EQE curves showed the increased A coefficient (from 8.3 × 10−6 s−1 in as-grown device to 1.0 × 10−7 s−1 in intermixed device) and decreased C coefficient (from 1.2 × 10−29 cm−3s−1 in as-grown device to 4.8 × 10−30 cm−3s−1 in intermixed device), corresponding to the SRH and Auger recombination process respectively. As the SRH recombination rate is proportional to n, while the Auger recombination rate is proportional to n3, the non-radiative recombination process is expected to be dominated by SRH recombination under low injection conditions and by Auger recombination at high injection currents. Therefore, the increased SRH recombination rate explains the lower peak EQE in potential energy engineered InGaN/GaN QW operating at relative low injection level. The possible reason for the increase is the introduction of vacancy like defects during metal/dielectric-induced interdiffusion. More importantly, the intermixed LED operating at high injection current showed a decrease in Auger recombination coefficient, which is consistent with simulation. The intermixed LED exhibited higher EQE than the as-grown device at J > 30 A/cm2. Thus the suppression of Auger related droop effect using intermixing provides a unique pathway towards higher performance LED devices.
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7997
Fig. 4. (a) EL spectra of as-grown, intermixed and non-intermixed LEDs. The peak energy values are indicated with the respective FWHM values indicated in brackets. (b) EQE vs. injection current density of as-grown, intermixed and non-intermixed InGaN/GaN QWs. Inset: voltage-current (V-I) relations of as-grown and intermixed device. The percentage of EQE droop at 150 A/cm2 is indicated.
4. Conclusion In summary, we demonstrated area-selective, and low temperature (950 °C) annealing technique based on metal/dielectric encapsulation intermixing technique for engineering the transition energy of InGaN/GaN QWs. A differential bandgap blueshift between SiO2 and metal/SiO2 encapsulated regions as large as 80 meV, corresponding to In diffusion length Ld of 10.9 Å at the QW interface, has been measured. We observed a significantly alleviated EQE droop of 14% in the intermixed QW LED, consistent with a reduction in RAuger with increasing Ld. The potential profile softened InGaN QWs after intermixing is effective in increasing the efficiency of c-plane oriented LEDs, desirable especially for high In composition structures, such as green LEDs. Our high spatial resolution, area-selective intermixing process is also applicable to monolithic-integration of photonic components in InGaN/GaN structures. Acknowledgments The authors gratefully acknowledge Yang Yang, Zaibin Guo, and Chao Zhao from Imaging and Characterization Corelab, KAUST, for assistance in PL set-up, RBS and HR-XRD measurements, respectively. The financial support King Abdulaziz City for Science and Technology (KACST) Technology Innovation Center (TIC) for Solid State Lighting at KAUST is gratefully acknowledged.
#234363 - $15.00 USD Received 11 Feb 2015; revised 11 Mar 2015; accepted 14 Mar 2015; published 19 Mar 2015 (C) 2015 OSA 23 Mar 2015 | Vol. 23, No. 6 | DOI:10.1364/OE.23.007991 | OPTICS EXPRESS 7998
