The semiconductor light-emitting diodes (LEDs) have the potential to be low-cost and long lifetime solid-state lighting sources [1–3]. And AlGaInP-based red LEDs have a strong current bearing capacity and high-temperature resistance which plays an important role in lighting, display, and traffic lights [4,5]. Many scientists and engineers are working to improve the output power and light extraction efficiency (LEE) [1,6–11], which reflects the ratio between the number of photons that emit outside the device and the number of photons that are produced. As we all know, the refractive index of the outside of the LED is lower than that of the inside LED, resulting in that most of the light emitted from semiconductor LEDs is reflected to the inner of the LED chip [12,13]. Even with 100% internal quantum efficiency (IQE), the LEE is limited to about 4% due to Fresnel reflection at the boundary [14,15]. In the past several decades years, various approaches have been proposed to enhance the LEE of LEDs [4,16–23].

Photonic crystals (PCs), which are materials with a spatially periodic refractive index, can control the direction of spontaneous emission, which is helpful to improve the LEE of LEDs [24,25]. LEDs with PCs (PCLEDs) could couple light from the dielectric-guided modes into the air to improve the LEE [26,27]. Much progress has been made in the fabrication and understanding of PCLEDs with emissions from the ultraviolet to the infrared region [24,27,28]. However, on the one hand, most of the experiments use electron beam lithography (EBL) or nanoimprinting to prepare nano-scale PCs, which is not conducive to rapid and repeated preparation [29,30]. On the other hand, most of the experiments made PC structures on the surface of the LED. And the best design is to put the active region in the middle of the PC [31]. In this article, PCLEDs with micro-scale PCs by normal ultra-violet lithography were fabricated. Furthermore, the active region of one sample was etched through the deep etching process, and then the active region was placed in the middle of the PC. Scanning electronic microscope (SEM) technology was used to characterize the morphology and cross-section picture. Photoluminescence (PL) spectrum was measured by grating spectrometer. Current-voltage (I–V) characteristics and light output power (LOP) were demonstrated by electroluminescent measurements.

In this work, AlGaInP-based red LED wafers were grown on lattice-matched GaAs (100) substrates with an electron concentration of 3×10¹⁸ cm^-3 by metal-organic chemical vapor deposition (MOCVD) system. The epi-layers were comprised of 6 pairs of (AlGa)_0.5In_0.5P/Ga_0.5In_0.5P MQWs with a total thickness of about 100 nm, following a 300-nm n-type InAlP cladding layer. Then a 900 nm p-type InAlP cladding layer and a 4-µm GaP window layer were grown. The epitaxial structure diagram can be seen in Fig. 1(a). Normal AlGaInP-based red LEDs (NLED), AlGaInP-based red LEDs with two dimensional (2D) surface photonic crystal (SPCLED), and AlGaInP-based red LEDs with 2D embedded photonic crystal (EPCLED) were fabricated as follows respectively, and the steps can be seen from Fig. 1(b). First, LED mesas of 1 mm×1 mm were formed by chemical etching after overlay photolithography. Then, as for NLED, Ti/Au top electrodes and Ni/Au/Ge/Ni/Au bottom electrodes were deposited by electron beam evaporation (EBE), followed by a rapid thermal annealing (RTA) process at 400 ℃ in N₂ ambient to improve the ohmic contact properties. As for SPCLED, the 2D dot array was made by normal UV lithography, followed by inductively coupled plasma reactive ion etching (ICP-RIE). The etching gas flow rates of BCl₃, Cl₂ and Ar are 10 sccm, 50 sccm, and 5 sccm respectively. The ICP power is 500 W, and the RF power is 35 W. The etching time is 8 min. And the etching depth is less than the thickness of GaP. Next, the same top electrodes and bottom electrodes were made. As for EPCLED, the 2D dot array was made by normal UV lithography, followed by ICP-RIE. The etching time is 20 min. And the etching depth exceeds the position of the quantum well. Next, we use plasma-enhanced chemical vapor deposition (PECVD) to deposit silicon dioxide (SiO₂) to fill the holes. Then, we use reactive ion etching (RIE) to remove SiO₂ and expose the GaP. Finally, the same top electrodes and bottom electrodes were made. Thus, we have prepared all the required samples.

 

Fig. 1. (a) The epitaxial structure diagram of AlGaInP-based red LEDs; (b) Preparation process steps of three samples.

Download Full Size | PDF

Next, field emission scanning electron microscope (SEM, Hitachi S-4800) was used to measure the surface morphology and cross-section picture with providing 10 kV high voltage to the electron gun. Furthermore, grating spectrometer was taken to measure the photoluminescence of all samples. As for measuring the electronic properties of LEDs, Keithley 4200 with the measurement step of 0.01 V was used. As for measuring the optical properties of LEDs, LED tester (LED-628) was taken to get the LOP with the measurement step of 2 mA. At the same time, pictures of the chips under low current were recorded.

SEM was used to study the surface morphology and cross-section picture of SPCLED and EPCLED respectively. Figure 2(a) and (b) show the SEM images of SPCLED. Figure 2(a) shows the surface picture of SPCLED and we can know that the diameter of the hole is 2 μm and the period of PC is 5 μm. Figure 2(b) shows the cross-section picture of SPCLED and we can know that the depth is 3 μm which is less than the thickness of GaP. Photonic crystals were fabricated on the surface. Figure 2(c)–(e) show the SEM images of EPCLED. Figure 2(c) shows the cross-section picture of EPCLED and we can know that the depth is 6 μm which is through the active layer, making the active region placed in the PC. But exposed active region would lead to a great current leakage and make a different to the electric properties. To improve this situation, SiO₂ was deposited as a passivation layer by PECVD. Figure 2(d) shows the surface picture of EPCLED after depositing SiO₂ by PECVD, from which we can see that SiO₂ has filled in the hole. Figure 2(e) shows the surface picture of EPCLED after etching SiO₂ on the surface by RIE to expose the surface of the GaP in order to form electrode contact. And we also can see that there still is SiO₂ on the walls of holes, which also can be proved by Fig. 2(f).

 

Fig. 2. (a) The SEM surface picture of SPCLED after etching by ICP. (b) The SEM cross-section picture of SPCLED etching by ICP. (c) The SEM cross-section picture of EPCLED etching by ICP. (d) The SEM surface picture of EPCLED after depositing SiO₂ by PECVD. (e) The SEM surface picture of EPCLED after etching SiO₂ by RIE. (f) The SEM cross-section picture of EPCLED after etching SiO₂ by RIE.

Download Full Size | PDF

 Figure 3 shows the photoluminescence testing of three LEDs. As we can see from Fig. 3(a), a 532 nm laser beam is emitted from the laser and then passed through a chopper and a converging lens to excite the samples. The excitation light of the samples enters the grating spectrometer through the collecting lens. The frequency of the chopper and the signal collected by the spectrometer were sent to the lock-in amplifier at the same time. Finally, the data were processed by professional software. In the testing process, three LEDs were put on the one wafer to make them in the same plane. By moving the base horizontally or vertically, the probe light is illuminated on different samples without changing any other parameters of the light path, which can make sure that all samples were excited under the same condition, which make the PL spectrum can be compared in value. From Fig. 3(e) we can see that all three samples’ photoluminescence peak is at 626 nm and the full width at half maxima (FWHM) is 10 nm for all three samples. Furthermore, the photoluminescence intensity of EPCLED is 60% higher than that of NLED and the photoluminescence intensity of SPCLED is 30% higher than that of NLED. An interesting phenomenon that different LEDs have different reflection patterns were found in the experiment and we can see that from Fig. 3(b)–(d). The reflection pattern of NLED was a bright point which is consistent with the spot of excitation light. However, the reflection patterns of SPCLED and EPCLED were period bright points array which can reflect the ability of control light in PC. Photon cannot escape from the material by multiple reflections or from the sidewall of chips, making that they only can escape from the hole in the PC. Because of this ability, stronger photoluminescence intensity from SPCLED and EPCLED can be got compared with NLED. And it also indicates that higher LOP and LEE can be got in the electroluminescence experiment.

 

Fig. 3. (a) Schematic diagram of photoluminescence testing equipment. (b) Reflection pattern of NLED. (c) Reflection pattern of SPCLED. (d) Reflection pattern of EPCLED. (e) Photoluminescence spectra of three samples.

Download Full Size | PDF

Electroluminescent properties of three LEDs are displayed in Fig. 4. I-V curves were obtained by Keithley 4200, and the LOP as a function of injection current is acquired by LED electron-optical tester. We can know that the electronic properties of the three LEDs are similarly good from Fig. 4(a). The electrical properties of the EPCLED decreased only a little compared with NLED and SPCLED. It demonstrates that depositing a passive layer of SiO₂ can repair defects of deep etching. And almost no defects are introduced into the epitaxial layer during the surface etching process. Figure 4(b) shows that LOP is improved by about 24% and 11% at 200 mA for SPCLED and EPCLED respectively, which is attributed to the suppression of the surface’s total reflection. The epitaxial layers of those three samples were grown at the same time and they have similar IQE and injection efficiency. Then we can say that the LEE has been improved by 24% and 11% at 200 mA respectively for SPCLED and EPCLED. The LOP of EPCLED starts to decay at around 190 mA and in the meantime, the LOP of NLED and SPCLED still keep increasing. Furthermore, the EPCLED has higher LEE than SPCLED at low current when it is less than 90 mA. It can be explained as follows. As is shown in Fig. 4(c), the output light of NLED is limited by Fresnel reflection. When the angle of the light is large enough, it is reflected into the chip. EPCLED is capable of controlling the light from the active region. By the photonic band gap effect and multiple reflection, EPCLED can guide most of the light along the direction perpendicular to the chip, which would enhance the LEE. SPCLED can control the light at the surface by the photonic band gap effect and diffraction, which would enhance the LEE too. So the LOP of EPCLED is larger than that of SPCLED at low injection current. Due to the fact that part of the active region is removed by etching, the current density under the same current is increased, making the heat generation seriously. As a result, EPCLED cannot work under higher current. Therefore, the LOP of SPCLED is larger than that of EPCLED at high injection current. Figure 4(d) shows the angle resolved EL spectra of the NLED, SPCLED, and EPCLED. The data has been normalized. EPCLED presents improved uniform light emission intensity in all directions among the three types of LEDs. The first line of Fig. 4(e) shows the digital pictures for three samples emitting light at 20 mA. And SPCLED and EPCLED have higher brightness and better uniformity compared with NLED. The second line of Fig. 4(e) shows the emission patterns of NLED, SPCLED, and EPLED at 1 mA. The black represents the electrode and the red part represents the electroluminescence. We can see that the electroluminescence of NLED is very weak and the electrode is completely black. As for SPCLED and EPCLED, the electroluminescence in the hole of photonic crystal is strong and can penetrate the electrode.

 

Fig. 4. (a) I-V curves of three samples. (b) The LOP of three samples. (c) Display of light outputs of NLED, SPCLED and EPCLED. (d) Angle resolved EL spectra of the NLED, SPCLED, and EPCLED. (e) The first line shows microscope pictures for three samples emitting light at 20 mA. The second line shows the emission patterns of NLED, SPCLED, and EPLED at 1 mA.

Download Full Size | PDF

LEDs with PC can get better LOP and have better uniformity, which is beneficial to Mini LEDs and Micro LEDs especially. On the one hand, enhancement of the LOP can ensure that enough light intensity can be got from small chips. On the other hand, uniform brightness is conducive to uniform color development and good imaging effect. And the microscale PC can be obtained by normal UV exposure easily.

In conclusion, high-brightness AlGaInP-based PCLEDs were fabricated by normal ultraviolet exposure combined with shallow etching and deep etching technique. SEM measurements show the surface pictures and cross-section pictures of PCLED in the preparation process. The photoluminescence intensity of SPCLED and EPCLED has been improved up to 60% and 30%, respectively, in comparison to NLEDs. And the reflection patterns of SPCLED and EPCLED were period bright points array, which shows the ability of control light in PC. Electroluminescent measurements show that three LEDs have a similar threshold voltage. The electrical properties of the EPCLED decreased slightly compared with SPCLED and NLED showing that depositing a passive layer of SiO₂ can repair defects of deep etching. Simultaneously, the light output power (LOP) of SPCLED and EPCLED has been improved up to 24% and 11% at 200 mA, respectively, in comparison to NLEDs. But the LOP decay earlier for EPCLED due to the excessive heat production. Furthermore, it is demonstrated that SPCLED and EPCLED luminous uniformity is better. This kind of high brightness PCLEDs can be widely used in Mini LEDs and Micro LEDs.