Effect of window layer with different growth modes on the photoelectric properties of AlGaInP LED

Quaternary AlGaInP-base light-emitting diode (LED) technologies have extensive applications in full-color displays, indoor lighting, traffic signals, and horticulture light due to the advantages of long life, energy saving, and low power consumption.¹ AlGaInP-base LED chip development has a mature structure framework,^2,3 but the low photoelectric efficiency remains the main issue limiting the application of AlGaInP-based LEDs.⁴ The photoelectric efficiency was strongly affected by the voltage and light output power. The light output power is mainly restricted by the mismatch of the refractive index between the materials of the device structure and ambient air, which leads to multiple total internal reflections at the interface. Meanwhile, the working voltage of the LED was affected by the series resistance of the LED and the Ohmic contact between the epitaxial layer and the electrode. Several methods have been proposed to increase the light extraction efficiency and decrease the working voltage.^5–7 Lee et al. proposed a chemical wet etching technology, using the method of adhesive layer bonding on n-side AlGaInP-based LEDs to form a triangular morphology of nano-rough surface to enhance light extraction efficiency and improve photoelectric efficiency.⁸ Tang et al. successfully prepared a high-brightness AlGaInP-based vertical LED with nanoscale silicon oxide (SiO₂) hemispherical arrays by photoresist thermal reflux technology and laser interference exposure to improve the photoelectric efficiency of LEDs.⁹ Choi et al. studied the development of Au/ITO Ohmic contacts with high transparency and low resistance for AlGaInP phosphor-based light emitting diodes on the p-GaP window layer to reduce the voltage.¹⁰ As an important structure of LED chips, the n-window layer shows an intense effect on spreading the current and improving light extraction efficiency.^11,12 However, up until now, there has been no detailed study about the effect of Al content in the n-window layer on the photoelectric performance of LEDs. The Al component in the AlGaInP system shows a great influence on its resistivity and refractive index and, furthermore, affects the light extraction efficiency and current spreading.^13,14 In this paper, several LEDs with different Al content growth modes in the n-AlGaInP window layer were designed and fabricated to improve the photoelectric efficiency of LED devices and the effect of Al content growth modes on the performance of LED devices was also discussed.

The AlGaInP red LEDs were grown on a Si-doped (100) GaAs substrate oriented 15° toward the ⟨111⟩ direction by an Aixtron low-pressure metal–organic chemical vapor deposition (MOCVD) G4 system. Trimethylaluminum (TMAl), trimethylgallium (TMGa), and trimethylindium (TMIn) were used as the group-Ⅲ sources. Phosphine (PH₃) and arsine (AsH₃) were used as the group-Ⅴ sources. Hydrogen (H₂) was used as the carrier gas. The LED structure consists of a GaAs contact layer, an n-AlGaInP window layer, an n-AlInP cladding layer, a GaInP/AlGaInP active region, a p-cladding layer, a GaP window layer, and a GaP contact layer. Three LED samples with different n-window growth modes were prepared according to the above process. As illustrated in Fig. 1, the n-window layer with gradient ridge Al content (Al content ranges from 0.15 to 0.45) was denoted as sample A. For a comparison, the n-window layer without with gradient ridge Al content (Al content is 0.15) was denoted as sample B and that without with gradient ridge Al content (Al content is 0.45) was denoted as sample C. After lithography, cleaning, evaporation, etching, and other processes, all LED samples were processed into chips with a size of 1 × 1 mm², and the detailed fabrication process is shown in Fig. 2.

Figure 3 shows the I–V curves of AlGaInP LEDs with different n-window growth modes. It can be found that the turn-on voltages of the three samples are approximately equal, but the slope (between 1.8 and 2.5 V) of sample C is smaller than that of samples A and B, indicating that the series resistance of sample C is greater than that of samples A and B. However, the slope of sample A is larger than that of samples B and C. The phenomenon can be attributed to the difference of ΔEc between the n-window layer and n-GaAs contact layer.^15–17 The Al content of sample C is higher than that of samples A and B; thus, the barrier difference between the n-AlGaInP window layer and the n-GaAs contact layer is also larger than that of samples A and B, and the series resistance of sample C is also the largest one among the three samples. Because the Al content of sample A at the interface is similar to that of sample B, the EC of the two is close. Meanwhile, although the slope of the I–V curve of sample A is almost equal to that of sample B, we found that the slope of sample A is a little larger than that of sample B. Figure 4 shows the current distribution model of the three samples in the N-type window layer. It can be seen that the current expansion performance of sample B is lower than that of sample A. Therefore, although the ΔEc of samples A and B is close, the slope of sample B is still smaller than that of sample A. This is attributed to the fact that compared with sample B, the growth mode of the n-type window layer of sample A is ridge gradient growth, which makes the parallel series resistance of the n-type window layer small, the lateral expansion ability of the current stronger, and the overall internal series resistance decrease.^18,19 Table I shows the photoelectric properties of the three samples under different working conditions. The working voltage of sample A is about 30 mV lower than that of sample B and 190 mV lower than that of sample C. It can be seen that the ridge gradient Al content in the n-window layer can effectively decrease the working voltage of the LED chips and the current crowding effect.²⁰ 

The dependence of light output power on the drive current for the AlGaInP LEDs is shown in Fig. 5(a), and Table II shows the light output power of the three LED samples at current densities of 24 and 110 A/cm². The light output power of LEDs in sample A is 290.1 mW at the current density of 24 A/cm², and it is 946.2 mW at the current density of 110 A/cm². The light output power of samples B and C increased from 279.1 to 276.6 mW at the current density of 24 A/cm² to 842.1 mW and 913.3 mW at 110 A/cm², respectively. At the same current density, the light output power of sample A is 12.4% and 3.6% higher than samples B and C, respectively. Therefore, the ridge gradient Al content can be used as a scheme to optimize the refractive index of the window layer.

According to total internal reflection, when the light enters a lower refractive index medium from a higher refractive index medium, total internal reflection will occur when the incident angle increases to the critical angle. The critical angle of total reflection is affected by the refractive index due to the different Al contents in the AlGaInP material. It can be obtained from Fig. 5(b) that the refractive index and extinction coefficient of AlGaInP decrease with the increase in the Al content. It can be obtained from Fig. 5(a) that the light output power of samples A, B, and C is the same at low current densities, which can be attributed to the low light output power at low current densities. When the current density gradually increases, the light output power difference between samples B and C gradually increases because with the current density increasing, the influence of the total reflection critical angle also increases. The Al content of sample A changes gradually, so the refractive index of the material also changes gradually, which makes the light deflect inside the AlGaInP material and changes the incident angle of the light at the interface between the n-AlGaInP window and the air. Therefore, the light that cannot be emitted at first can enter the air after deflecting inside, resulting in that the light output power of sample A will be higher than that of samples B and C.

TracePro software is used to simulate the propagation path and light output power in three n-window layers with different Al contents. The n-window layer is optically simulated using a 20 W light source, and the number of rays is set to 150. When the number of rays is too large, the propagation path of light cannot be clearly observed, so the number of rays of 150 is selected for the simulation. Figures 6(a)–6(c) show the simulation results of the propagation of light in the n-type window layer of different Al contents. As can be seen from these figures, the number of emitting lights by sample A is larger than that of samples B and C. At the same time, it can be found in Table III that the maximum light output power of sample A is 1181.2 mW; for sample C, the maximum is about 918.5 mW; and for sample B, it is about 813.2 mW. The trend of our experimental results is the same as that of the simulation results.

At the same time, there is also a factor that causes the light output power of the three samples to be different. It can be seen from Fig. 5(a) that the lateral current expansion ability of sample A is stronger than that of samples B and C, so the current injection area of sample A is larger than that of the other two samples and the light output power is larger than that of the other two samples. Similarly, the light output power of sample B will be higher than that of sample C. In summary, the light output power of the three samples is different because of the refractive index of the material and the area of the active region of the current injection.

The wall-plug efficiency (WPE) is used to characterize the photoelectric conversion efficiency of LEDs. The calculation is based on the following equation:

where Φ is the light output power and V_f and I_f are the forward working voltage and current of the LED, respectively.²¹ The n-type window layer design of sample A changes the barrier difference between the n-window and n-contact, which reduces the working voltage. At the same time, the design also changes the refractive index of the n-window, which increases the output power of light at high current densities. Under rated working conditions, the WPE of samples A, B, and C is about 48.86%, 46.41%, and 43.19%, respectively. Compared with that of samples B and C, the WPE of sample A is increased by about 2.45% (Al = 0.15) and 5.68% (Al = 0.45), respectively.

Figure 7 shows the WPE vs current density curve of the three samples under different forward currents. According to the results of Tseng et al.,²² the decline rate of the WPE can be characterized by the droop efficiency, which is defined as [(WPE_max − WPE_min)/WPE_max] × 100%.

In the above definitions, WPE_max and WPE_min are the maximum and minimum photoelectric efficiencies, respectively. Table IV shows the WPE_max, WPE_min, and droop efficiency of the three samples. The WPE_max of samples A, B, and C is about 58.03%, 55.96%, and 55.87%, respectively, and the WPE_min of the three samples is about 23.51%, 20.38%, and 18.80%, respectively. The droop efficiency value of the three samples is about 65.53%, 74.98%, and 78.52%, respectively, indicating that the droop efficiency of sample A is slower than that of samples B and C. Due to the improvement of the luminous efficiency and the decrease in the forward working voltage of LED, the WPE of sample A is the highest one among the three samples. Meanwhile, the internal resistance of sample A is less than that of the other two samples.

In this paper, n-AlGaInP window layer red LEDs with three different Al content growth modes were prepared to investigate the effect of Al content in the n-AlGaInP window layer on the photoelectric performance of LEDs. The results show that the forward voltage of red LEDs with the (Al_xGa_1−x)_0.5In_0.5P n-window layer with the ridge gradient Al content was 30 mV (x = 0.15) and 190 mV (x = 0.45), respectively, which are lower than that of the n-window layer without gradient ridge Al content samples. The light output power was 12.3% (x = 0.15) and 3.6% (x = 0.45) higher than that of the samples with gradient ridge Al component, and the photoelectric efficiency was about 2.45% (x = 0.15) and 3.6% (x = 0.45) higher. The efficiency drop effect also is slower than the other samples. The red LEDs with the n-AlGaInP window layer with the ridge gradient Al content can effectively reduce the internal resistance of LEDs, enhance the current expansion ability of the n-window layer, and also improve the light output power of LEDs because the Al component of the n-window layer is ridge gradient. It can be found that the ridge gradient Al in the n-window layer can improve the luminous efficiency of the red plant lighting LEDs and they are more conducive to plant factories to replace traditional light sources.

This work was supported by the Fujian Provincial Natural Science Foundation of China (Grant No. 2020J01297).

The authors have no conflicts to disclose.

Xie Lanchi wrote the manuscript and participated in all the experiments and the data analysis. Xiong Feibing, Li Senlin and Bi Jingfeng were involved in the writing of the manuscript. Xue Long and Li Senlin participated in all the experiments and the data analysis. Wang Yahong, Lai Yucai, Liao Yinsheng, Yang Meijia took part in the discussions and testing of photoelectric performance. Xiong Feibing and Li Senlin supervised the writing of the manuscript and all the experiments. Bo Wang participated in simulation work. All authors read and approved the final manuscript.

Lanchi Xie: Data curation (equal); Formal analysis (equal). Senlin Li: Investigation (equal); Methodology (equal); Supervision (equal). Jingfeng Bi: Resources (equal); Software (equal); Writing – original draft (equal). Long Xue: Resources (equal); Validation (equal); Writing – original draft (equal). Yahong Wang: Data curation (equal); Investigation (equal); Resources (equal). Yucai Lai: Data curation (equal); Formal analysis (equal). Yinsheng Liao: Formal analysis (equal); Methodology (equal). Xuezhen Dong: Data curation (equal); Investigation (equal). Meijia Yang: Data curation (equal); Formal analysis (equal). Bo Wang: Data curation (equal); Software (equal). Feibing Xiong: Funding acquisition (lead); Supervision (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.