Polarization-reversed electron-blocking structure, which had negative polarization charges localized at the interface between the last quantum barrier (LQB) and electron-blocking layer (EBL), was demonstrated to remarkably improve the light-emitting efficiency of GaN-based blue light-emitting diodes (LEDs) numerically and experimentally. The improvement was attributed to the enhanced electron-blocking effectiveness by the elevated conduction band nearby the LQB/EBL interface. Nevertheless, the efficiency droop was not mitigated because the decrease of electron-leakage was accompanied by the increase of Auger recombination.
© 2014 Optical Society of America
Highly efficient GaN-based light-emitting diodes (LEDs) have been intensively developed in recent years for various applications, such as full color displays, liquid crystal display back lightings and solid-state lighting. Though great success has been achieved in improving the LEDs performance, it is still desired to further increase the internal quantum efficiency (IQE) and suppress the efficiency droop under high injection levels to make LEDs more energy-efficient and cost-competitive than conventional lighting technologies. One critical factor that limits the IQE and leads to the efficiency droop is electron-leakage, originating from the injection mismatch of electrons and holes and further aggravated by the polarization effect in conventional LEDs made up of Ga-plane materials [1–3]. To reduce the electron-leakage, one general approach is inserting a wide-bandgap AlGaN electron blocking layer (EBL) between the last quantum barrier (LQB) and p-region to prevent electrons overflowing out of the active region. However, this approach is subject to the positive polarization charges localized at the interface between LQB and AlGaN EBL. The positive charges can drop the conduction band edge nearby the LQB/EBL interface and thus they are unfavorable for blocking electron leakage. It doesn’t help even if we increase the Al composition in AlGaN EBL, because more polarization charges and larger band drop will be induced. Great efforts were made to mitigate the electron-leakage. Among these, many efforts were made to improve the effectiveness of EBL by reducing the polarization effect, including grading or tapered EBL [4–7], lattice-matched AlInN EBL , polarization-matched AlInGaN EBL [1, 9, 10], wafer- or chip-bonding process to invert EBL polarization . Besides, several strategies were proposed to impede the electron injection and simultaneously enhance the hole injection, including p-type InGaN inserted between LQB and EBL , graded LQB , and asymmetric multi-quantum wells . Stair-case electron injector concept was suggested as an “electron cooler” to reduce electron-overflow [15, 16]. In this work, we propose to utilize the polarization to suppress the electron-leakage by using polarization-reversed electron-blocking structure with negative polarization charges. It demonstrates superior performance numerically and experimentally. It has the same target to induce negative polarization charges as the bonding strategy proposed by Meyaard et al . Nevertheless, the bonding strategy requires two separate epitaxial growth followed by wafer- or chip- bonding and laser lift-off, in which the bonding process is a particular challenge considering the commonly existed wafer-bending. Relatively speaking, the proposed structure in this paper is rather simple and flexible for material growth.
2. Theoretical analysis
Using conventional AlGaN EBL, positive polarization charges at LQB/EBL interface caused detrimental effect. Polarization-matched AlInGaN EBL can effectively solve the problem, but it required quarternary AlInGaN EBL with strict alloy composition which was rather difficult for material growth. How would it evolve if the interface polarization charges are negative? To investigate the effect, simulation on a same GaN-based blue LED with different LQB/EBL interface polarization charges was performed using APSYS modeling software . The LED consisted of n-GaN, eight periods In0.21Ga0.79N/GaN quantum wells, 12 nm GaN LQB, 40 nm Al0.1In0.047Ga0.853N EBL, and 80 nm p-GaN. The thickness for quantum wells and quantum barriers were 2.5 nm and 12 nm, respectively. The polarization and bandgap of the quaternary AlInGaN alloy were determined by the model given in . Commonly accepted parameters , including an Auger coefficient of 2 × 10−30 cm6s−1 and conduction band offset ratio of 60% were adopted [19–21]. The Al0.1In0.047Ga0.853N EBL should be polarization-matched to the GaN LQB, however, in order to explore the effect of interface polarization charges, simulation with assumed polarization charges of + 1.8 × 1016 m−2, zero, and −1.8 × 1016 m−2 was carried out respectively. The value of + 1.8 × 1016 m−2 was on the same order of magnitude as the interface polarization in the conventional GaN LQB/AlGaN EBL structure, which would be described later. The calculated band diagrams were illustrated in Fig. 1(a). It was important to note that the conduction band nearby LQB/EBL interface displayed sharp contrast. In case of positive polarization charges, it was dramatically dropped and unfavorable for blocking electrons leakage. In case of zero polarization charges, it was basically flat as reported . In case of negative polarization charges, it was elevated and thus more beneficial to suppress electron-leakage. With positive, zero and negative polarization charges, the effective energy barrier height for electrons (En) increased successively from 275 meV, 384 meV to 492 meV. En in case of negative polarization charges were contributed by both the LQB/EBL conduction band offset and the upward-bending LQB band. High effective barrier height was of significant benefit to depress electrons leakage. Correspondingly, the electron density in p-region, as illustrated in Fig. 1(b), decreased successively. The parasitic electrons at LQB/EBL interface in case of positive charges cannot contribute to the desired light-emission, but was subject to nonradiative recombination. In cases of zero and negative charges, the conduction band dip at LQB/EBL interface was removed and the electron accumulation at LQB/EBL interface was effectively eliminated. Besides, the nonhomogeneous distribution of electrons in MQW was also improved from positive, zero to negative polarization charges, as shown in Fig. 1(c). The electrons density decreased in the first QW nearest to p-region, while increased in the subsequent QWs. This could also improve the light-emission efficiency. So, devising negative polarization charges at the interface between LQB and EBL, seemed a good tactic to improve the LED efficiency.
To induce negative polarization charges at LQB/EBL interface, however, is difficult to realize for LEDs based on Ga-plane III-nitrides. EBL with wider bandgap relative to LQB was generally desired, but it would induce positive polarization charges because broad-bandgap corresponds closely to strong-polarization for III-nitrides, both basically linearly-depending on the lattice constant. For this reason, a simple structure using AlGaN in place of LQB and GaN in place of EBL was considered, which had negative interface polarization charges, although the bandgap of the nominal GaN EBL was smaller than AlGaN LQB. Here, it should be noted that the nominal GaN EBL itself cannot block electrons. It was labeled as EBL just for ease of presentation. This structure and several reference structures with different LQB/EBL materials collocations, as listed in Table 1, were also simulated by APSYS. Sample A had a conventional structure using GaN as LQB and Al0.15Ga0.85N as EBL, which had positive polarization charge localized at LQB/EBL interface. Sample B used GaN as LQB and polarization-matched Al0.1In0.047Ga0.853N as EBL. Sample C used Al0.15Ga0.85N as LQB and GaN as nominal EBL, which is the proposed structure that had negative polarization charge localized at LQB/EBL interface. Sample D used Al0.15Ga0.85N as LQB and polarization-matched Al0.2In0.026Ga0.774N as EBL. The LQB/EBL interface polarization charges should be + 0.01 C/m2 for sample A if the materials were pseudomorphically grown on relaxed GaN, but they could be partially compensated by the stain relaxation, the charged defects and the injected carriers. The exact value of the remaining polarization is unknown. In the simulation, only 30%, i.e. + 1.8 × 1016 m−2 remaining polarization charge was adopted. Same amount of negative polarization charges was adopted for sample C. No polarization charges existed for sample B and sample D. The conduction band offset (ΔEC) was also listed in Table 1. The calculated band diagrams were compared in Fig. 2. It should be noticed, for sample C, electrons were blocked by the rising band of LQB itself instead of the conduction band offset between LQB and EBL. The effective potential barrier height for electrons gradually increased from sample A (264 meV), sample B (329 meV) to sample C (465 meV). The high potential barrier for electrons in sample C was of significant benefit to depress electrons leakage. Nevertheless, it should be noted that the rising LQB band, which contributed to the high barrier for electrons, also resulted in a high barrier for holes. As for sample D, it also had polarization-matched LQB/EBL as sample B, but its bandgap of both LQB and EBL was higher and thus more favorable for blocking electron-leakage. Compared to sample C, sample D also showed higher barrier for electrons together with lower barrier for holes by virtue of its wie-bandgap EBL. From this respect, sample D was better than sample C.
3. Experimental results
To testify the concept experimentally, the four GaN-based blue LED samples referred above were grown by one commercial Veeco metal-organic chemical vapor deposition (MOCVD) system on c-plane sapphire substrates. Ammonia (NH3), Triethylgallium (TEGa), trimethyl-indium (TMIn) and trimethylaluminum (TMAl) were used as source materials. After the deposition of 25 nm thick GaN nucleation layer under 550 °C followed by 2 µm undoped GaN layer and 2 µm Si-doped GaN layer under 1050 °C, the InGaN/GaN MQW was deposited. The GaN barriers were deposited under 830 °C and unintentionally doped. The InGaN wells were deposited under 730 °C with the flow rate of 15 sccm for TEGa, 160 sccm for TMIn, and 6.6 slm for NH3 respectively. Subsequently, LQB and EBL were deposited. LQBs of all samples were deposited under 830 °C and unintentionally doped. EBL of sample A was deposited under 950 °C. EBL of sample B was deposited under 805 °C with respective flow rate of 44 sccm for TEGa, 260 sccm for TMIn, 44 sccm for TMAl and 6.6 slm for NH3. EBL of sample C was also deposited under 950 °C. EBL of sample D was deposited under 830 °C with same flow rate as EBL of sample B. EBLs of all samples were Mg-doped with holes density of about 1 × 1017 cm−3. Finally, p-type GaN with holes density of about 3 × 1017 cm−3 was deposited under 950 °C to complete the whole epitaxial structure. The wafers were then processed to form 1 × 1 mm2 LED chips using a conventional mesa structure method and encapsulated for device characterization. The electrical and luminescence characteristics of the four LEDs were measured with a calibrated integrating sphere, including the current-voltage (I-V), the light-emission spectra and the light output power-current (L-I).
The forward I-V characteristics, as shown in Fig. 3(a), revealed that the forward voltage (VF) of sample C (3.35 V) were smaller than sample A (3.45 V) and sample D (3.59 V) under the common operation current of 350 mA, but somewhat larger than sample B (3.22 V). The higher measured value of VF in sample D might be caused by the high-resistance of the AlInGaN layer due to the difficulty in magnesium-doping. The leakage current, as shown in Fig. 3(b), was 0.16 μA, 0.76 μA, 0.16 μA and 0.26 μA for sample A-D respectively under reverse-bias of 10 V. Sample C had lower leakage than sample B and sample D. The EL spectra, measured under the same injection current of 350 mA, were depicted in Fig. 3(c). Remarkably, sample C showed the strongest light-emission intensity. Compared to sample A, the output power was enhanced by 9% for sample B, 25% for sample C, and 17% for sample D. Sample D showed higher intensity than sample B, which could be attributed to the reduced electron-leakage due to the higher-bandgap LQB/EBL and the higher-quality AlInGaN EBL due to the higher growth temperature. Nevertheless, sample D showed lower intensity than sample C. Considering the difficulty in AlInGaN material growth and the harsh requirement on the alloy composition for polarization-matching, the efficiency of sample D was probably degraded by its poor material quality or deviated alloy composition. Maybe sample D could behave better than sample C by optimizing material growth, but it should be also noticed sample C suffered from the low-bandgap EBL. Additionally, the peak wavelength of sample B-D were about 7 nm blue shifted compared with sample A. It was speculated that the polarization field in the wells, especially the QW nearest to p-region (LQW), could be reduced to some extent by removing the positive polarization charges, which could alleviate the quantum-confined Stark effect and therefore contribute to the blue shift. Besides, when AlGaN LQB was used in place of GaN LQB, the quantum confinement effect in LQW could be enhanced due to the increased band-offset at the LQW/LQB interface, which can also cause the blue shift. The light output power versus injection current for these LEDs, measured in a continuous-wave current mode, was presented in Fig. 3(d). Basically, the output power gradually increased from sample A, sample B, sample D to sample C under the overall current range from 5 mA to 1 A.
So, the experimental data testified that designing negative polarization charges in electron-blocking structure was an advantageous approach to improve the light-emission efficiency. In comparison with the polarization-matching approach, this approach was much easier and more flexible from the view of material growth. Polarization-matching required quarternary AlInGaN EBL with strict alloy composition, which was rather difficult to grow. Poor AlInGaN material quality and deviated alloy composition might deteriorate the device performance, resulting in the relative lower output power of sample D than that of sample C.
4. Efficiency droop analysis
The efficiency droop, which refers to the decrease of light-emission efficiency under high electrical injection levels, is a hot topic attracting great attention. Its physical origin is still under debate. Proposed mechanisms include electron-leakage , Auger recombination , and carrier delocalization , etc. The droop behavior was also probed for our samples. Figure 4(a) represented the EQE versus injection current. Although sample C demonstrated reduced electron-leakage and improved efficiency, it was surprising that it did not show some mitigation of efficiency droop. It was suspected, although the reduced electron-leakage should result in reduced droop, it could also lead to higher carrier density in MQW and thus more Auger recombination loss, which could cause additional efficiency droop to counteract or even exceed the reduced droop amount contributed by the enhanced electron-blocking effectiveness. This speculation was supported by simulation. As shown in Fig. 4(b), simulation revealed that the electron-leakage was eliminated virtually but accompanied by the remarkable increase of Auger recombination loss. The IQE was largely limited by electron-leakage for sample A, while critically limited by Auger loss for sample C.
In conclusion, GaN-based blue LEDs with different polarization charges in electron-blocking structure have been investigated both numerically and experimentally. Polarization-reversion approach, which had negative polarization on the LQB/EBL interface, demonstrated superior light-emission efficiency in comparison with the conventional AlGaN EBL approach. It also showed advantages over the polarization-matching approach. It exhibited higher efficiency experimentally and it was much easier and more flexible for material growth. This approach was very simple but highly effective. It could be employed into various polar GaN-based LEDs or laser diodes.
This work was supported the National Natural Sciences Foundation of China under Grant 61274040, the National High Technology Program of China under Grant 2014AA032605 and the National Basic Research Program of China under Grant 2011CB301902.
References and links
1. M.-H. Kim, M. F. Schubert, Q. Dai, J. K. Kim, E. F. Schubert, J. Piprek, and Y. Park, “Origin of efficiency droop in GaN-based light-emitting diodes,” Appl. Phys. Lett. 91(18), 183507 (2007). [CrossRef]
2. K. J. Vampola, M. Iza, S. Keller, S. P. DenBaars, and S. Nakamura, “Measurement of electron overflow in 450 nm InGaN light-emitting diode structures,” Appl. Phys. Lett. 94(6), 061116 (2009). [CrossRef]
3. D. S. Meyaard, G. Lin, Q. Shan, J. Cho, E. F. Schubert, H. Shim, M.-H. Kim, and C. Sone, “Asymmetry of carrier transport leading to efficiency droop in GaInN based light-emitting diodes,” Appl. Phys. Lett. 99(25), 251115 (2011). [CrossRef]
4. J.-H. Ryou, P. D. Yoder, J. Liu, Z. Lochner, H. Kim, S. Choi, H. J. Kim, and R. D. Dupuis, “Control of Quantum-Confined Stark Effect in InGaN-Based Quantum Wells,” IEEE J. Sel. Top. Quantum Electron. 15(4), 1080–1091 (2009). [CrossRef]
5. Y.-K. Kuo, J.-Y. Chang, and M.-C. Tsai, “Enhancement in hole-injection efficiency of blue InGaN light-emitting diodes from reduced polarization by some specific designs for the electron blocking layer,” Opt. Lett. 35(19), 3285–3287 (2010). [CrossRef] [PubMed]
6. N. Zhang, Z. Liu, T. Wei, L. Zhang, X. Wei, X. Wang, H. Lu, J. Li, and J. Wang, “Effect of the graded electron blocking layer on the emission properties of GaN-based green light-emitting diodes,” Appl. Phys. Lett. 100(5), 053504 (2012). [CrossRef]
7. C. H. Wang, C. C. Ke, C. Y. Lee, S. P. Chang, W. T. Chang, J. C. Li, Z. Y. Li, H. C. Yang, H. C. Kuo, T. C. Lu, and S. C. Wang, “Hole injection and efficiency droop improvement in InGaN/GaN light emitting diodes by band-engineered electron blocking layer,” Appl. Phys. Lett. 97(26), 261103 (2010). [CrossRef]
8. S. Choi, H. J. Kim, S.-S. Kim, J. Liu, J. Kim, J.-H. Ryou, R. D. Dupuis, A. M. Fischer, and F. A. Ponce, “Improvement of peak quantum efficiency and efficiency droop in III-nitride visible light-emitting diodes with an InAlN electron-blocking layer,” Appl. Phys. Lett. 96(22), 221105 (2010). [CrossRef]
9. M. F. Schubert, J. Xu, J. K. Kim, E. F. Schubert, M. H. Kim, S. Yoon, S. M. Lee, C. Sone, T. Sakong, and Y. Park, “Polarization-matched GaInN/AlGaInN multi-quantum-well light-emitting diodes with reduced efficiency droop,” Appl. Phys. Lett. 93(4), 041102 (2008).
10. A. J. Ghazai, S. M. Thahab, H. Abu Hassan, and Z. Hassan, “Quaternary ultraviolet AlInGaN MQW laser diode performance using quaternary AlInGaN electron blocking layer,” Opt. Express 19(10), 9245–9254 (2011). [CrossRef] [PubMed]
11. D. S. Meyaard, G.-B. Lin, M. Ma, J. Cho, E. F. Schubert, S.-H. Han, M. H. Kim, H. W. Shim, and Y. S. Kim, “GaInN light-emitting diodes using separate epitaxial growth for the p-type region to attain polarization-inverted electron-blocking layer, reduced electron leakage, and improved hole injection,” Appl. Phys. Lett. 103(20), 201112 (2013). [CrossRef]
12. R. M. Lin, S. F. Yu, S. J. Chang, T. H. Chiang, S. P. Chang, and C. H. Chen, “Inserting a p-InGaN layer before the p-AlGaN electron blocking layer suppresses efficiency droop in InGaN-based light-emitting diodes,” Appl. Phys. Lett. 101(8), 081120 (2012). [CrossRef]
13. C. S. Xia, Z. M. S. Li, W. Lu, Z. H. Zhang, Y. Sheng, and L. W. Cheng, “Droop improvement in blue InGaN/GaN multiple quantum well light-emitting diodes with indium graded last barrier,” Appl. Phys. Lett. 99(23), 233501 (2011). [CrossRef]
14. J. Y. Zhang, L. E. Cai, B. P. Zhang, X. L. Hu, F. Jiang, J. Z. Yu, and Q. M. Wang, “Efficient hole transport in asymmetric coupled InGaN multiple quantum wells,” Appl. Phys. Lett. 95(16), 161110 (2009). [CrossRef]
15. X. Ni, X. Li, J. Lee, S. Liu, V. Avrutin, Ü. Özgür, H. Morkoç, A. Matulionis, T. Paskova, G. Mulholland, and K. R. Evans, “InGaN staircase electron injector for reduction of electron overflow in InGaN light emitting diodes,” Appl. Phys. Lett. 97(3), 031110 (2010). [CrossRef]
16. V. Avrutin, S. A. Hafiz, F. Zhang, Ü. Ozgür, H. Morkoç, and A. Matulionis, “InGaN light-emitting diodes: Efficiency-limiting processes at high injection,” J. Vac. Sci. Technol. A 31(5), 050809 (2013). [CrossRef]
17. APSYS Device Simulator, Software Package, Crosslight Software, Inc., Canada.
18. I. Vurgaftman and J. R. Meyer, “Band parameters for nitrogen-containing semiconductors,” J. Appl. Phys. 94(6), 3675 (2003). [CrossRef]
19. Y. Shen, G. Mueller, S. Watanabe, N. Gardner, A. Munkholm, and M. Krames, “Auger recombination in InGaN measured by photoluminescence,” Appl. Phys. Lett. 91(14), 141101 (2007). [CrossRef]
20. K. T. Delaney, P. Rinke, and C. G. Van de Walle, “Auger recombination rates in nitrides from first principles,” Appl. Phys. Lett. 94(19), 191109 (2009). [CrossRef]
21. J. Piprek and Z. M. S. Li, “Sensitivity analysis of electron leakage in III-nitride light-emitting diodes,” Appl. Phys. Lett. 102(13), 131103 (2013). [CrossRef]
22. 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). [CrossRef] [PubMed]
23. X. Cao, Y. Yang, and H. Guo, “On the origin of efficiency roll-off in InGaN-based light-emitting diodes,” J. Appl. Phys. 104(9), 093108 (2008). [CrossRef]