Open Access
Add to BibSonomy
Abstract
This study fabricated high-voltage, low-current DUV-LEDs by connecting two devices. Due to better current spreading and the enhanced reflective mirror effect, high-voltage devices present a higher dynamic resistance, emission output power, wall-plug efficiency, external quantum efficiency, and view angle than single traditional devices. The study found that when the injection current was 320 mA, the maximum output power was exhibited at 47.1 mW in the HV sample. The maximum WPE and EQE of high-voltage DUV-LEDs were 2.46% and 5.48%, respectively. Noteworthily, the redshift wavelength shifted from 287.5 to 280.5 nm, less than the traditional device—from 278 to 282 nm. Further, due to the uniform emission patterns in high-voltage devices, the view angle presents 130 degrees at 100 mA input current. In this study, the high-voltage device showed more excellent properties than the traditional device. In particular, it presented a high potential application in high-voltage circuits, which can remove transformers to eliminate extra power consumption.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
The COVID-19 virus has been widespread since 2020, resulting in the shortage of many disinfection products made of ultraviolet (UV) light. Deep ultraviolet light-emitting diodes (DUV-LEDs) as a light source with wavelengths of 220-350 nm (5.6-3.5eV) are expected to be used in a wide variety of fields, such as bio/chemical devices, medical diagnostics treatment, environmental protection, sterilization, water purification, phototherapy, and public health [1]. The bandgap energy of aluminum gallium nitride (AlGaN) covers from 3.4 eV of GaN to 6.2 eV of AlN, making the AlGaN-based semiconductor the most promising material of DUV-LEDs. For different applications, the emission wavelength can be adjusted by the aluminum content in the AlGaN epi-layer structure [2]. Over the last decade, many researchers presented several studies on high-efficiency DUV LEDs [1,3–6] to satisfy the potential applications mentioned above. DUV-LEDs offer significant advantages such as small chip size, low operation voltage, spectra tunability, long lifetime, and good temperature stability, which can substitute the traditional mercury vapor lamps [7]. Most importantly, DUV-LEDs are environment-friendly.
However, DUV-LEDs have usually grown on sapphire substrates by the heteroepitaxial method, which has considerable lattice mismatch, dislocation densities, and material defects. To improve the device performance, optimizing the growth parameters in DUV-LEDs is necessary [8]. Past reports indicated that compared to blue LEDs, DUV-LEDs have serious issues, such as lattice mismatch between the sapphire substrate and AlGaN epi-layers. The number of defects and dislocations in active regions introduces lower output power, resulting in lower internal quantum efficiency (IQE) and external quantum efficiency (EQE) [9,10]. The light extraction efficiency (LEE) is also extremely low due to the absorption of the p-type GaN layer in the DUV-LEDs structure. Further, n-AlN and sapphire reflect the light to p-GaN, which is absorbed by the p-GaN layers and reduces the optical property of DUV-LEDs. In 2021, the Y. Matsukura group exhibited relatively high EQE (15.7% at 275 nm) in DUV-LEDs by designing a high-Al-composition p-AlGaN cladding layer, thin p-GaN contact layer, and reflective p-type electrode structures to reduce the light absorption in the p-GaN layer and enhance the LEE. Notably, the output power increased from 150 mW to 330 mW with an injection current of 1 A [11]. In the same year, S. H. Lin et al. presented the reflective passivation layer as Bragg reflectors, which consists of HfO2 and SiO2 stack layers [12]. The EQE and emission output power were enhanced by the stacking layers, and the maximum wall plug efficiency (WPE) was improved from 2.01% to 2.59%. Meanwhile, many researchers discuss the packaging technologies of DUV-LEDs to improve their properties [13–16].
Given the underperformance caused by the epitaxial process and epi-layer structure, fabricating products with DUV-LEDs always needs to be integrated with the circuits and merged with the electronic components. In general, DUV-LEDs are always operated at low voltage. It is usually necessary to reduce the input voltage by an extra transformer as the LED driver in the components of the circuits, increasing the driver volume and obtaining additional power consumed. Therefore, a high voltage (HV) DUV-LED was designed and examined in this paper. The HV device structure can remove a transformer in the circuit and prevent the optical properties from decreasing. This study investigated the characteristics of HV DUV-LEDs operating with low current injection, including current-voltage (I-V) characteristics, emission output power, EQE, view angle, and operating temperature. A low-voltage traditional DUV-LED power chip operated at low voltage was prepared for comparison.
2. Experimental
The AlGaN-based structures of DUV-LEDs were grown on (0001) sapphire substrate with a 25 nm AlN template layer by metalorganic chemical vapor deposition (MOCVD). Disilane (Si2H6) and bis (cyclopentadienyl) magnesium (Cp2Mg) were used for n- and p-type doping, respectively. Trimethylgallium (TMGa) and trimethylaluminum (TMAl) were used for the group-III element, while ammonia (NH3) was employed for the group-V element. A detailed structure has been published in our previous work. It consists of an AlN layer with 1.5 μm-thickness, an Si-doped Al0.45Ga0.55N layer with 2 μm-thickness, an Si-doped Al0.4Ga0.6N/Al0.6Ga0.4N multiple quantum well (MQW), a p-Al0.75Ga0.25N as an electron-blocking layer (EBL) with 30 nm-thickness, a p-GaN:Mg layer, and an Mg-doped p+-GaN of a 5 nm-thick contact layer. The total thickness of p-GaN and p+-GaN in this study were 100 nm and 40 nm, respectively. After the epi-layer growth, the sapphire was polished to reduce the thickness of the substrate for the subsequent dicing process, as shown in Fig. 1(a). Then, the inductively coupled plasma reactive ion etching (ICP-RIE) was used to etch the active region for the exposure of the n-AlGaN layer, and the semiconductor was isolated, as shown in Fig. 1(b). Subsequently, the metal and ITO were deposited by e-gun evaporation for the contact pads, as presented in Fig. 1(c). In order to obtain a high voltage (HV) in DUV-LEDs, two series chips were designed from the first chip n-side to the second chip p-side. In this step, the current blocking (CB) material was covered on the sidewall of the first chip to isolate the cathode and anode, then the metal bridge between two chips was deposited on CB and two contact chips in series connection, as shown in Fig. 1(d). Figs. (e) and (f) show the dielectric passivation layer and bonding pad, which were deposited by plasma-enhanced chemical vapor deposition (PECVD) and thermal evaporation, respectively. Finally, the HV DUV-LED was diced and packaged on an AlN substrate with metal pads by flip-chip bonding, as shown in Fig. 1(g). Figure 1(h) shows the top and bottom views of HV chips measured by an optical microscope. For comparison, the ST DUV-LEDs were also prepared. The corresponding top and bottom views of HV chips measured by the optical microscope are shown in Fig. 1(i). The ST and HV chip sizes of these two samples were 20 × 20 mil2 and 10× 20 mil2 connected to 10 × 20 mil2, respectively. It must be noted that the active areas of ST and HV DUV-LEDs were maintained at 400 mil2.
In comparing the characteristics of ST and HV DUV-LEDs, the electrical and optical properties of two DUV-LEDs were measured at room temperature. The electrical characteristic of the current-voltage (I-V) curve was measured using a semiconductor parameter analyzer (Keithley, 2400 Source Meter). The optical output power was measured by a calibrated integrating sphere after the flip chip bonding process, and the far-field radiation patterns between ST and HV samples were measured using an integrating sphere detector (CAS 140B, Instrument Systems). To better illustrate the performance of the device, the WPE and EQE were introduced and calculated as follows:
3. Result and discussion
Figure 2 (a) presented the current as function of voltage and dynamic resistance. The turn on voltages were 5 V and 10 V for ST and HV DUV LEDs, respectively. Obviously, the higher forward voltage of DUV-LEDs was obtained successfully by a series connecting two 10×20 mil2 chips. The turn on voltage of HV DUV LED was twice times of that ST DUV LEDs due to two small DUV LEDs series. The dynamic resistances of ST and HV at injection current density 120 A/cm2 (320 mA for ST and 160 mA for HV) were 2.5 Ω and 5.9 Ω, respectively. It shows that the resistance of HV DUV-LEDs was 2 times higher than that of traditional ST DUV-LEDs. Figure 2(b) shows the voltage as the function of the current (V-I) curves of these two DUV-LEDs, indicating that the voltage of the HV chip was almost twice as high as the ST sample at the same injection current. When the input current was 1 mA, the forward voltages were 9.9 V and 4.8 V for the HV and ST DUV-LEDs, respectively. The voltage of the HV sample was two times more than that of ST DUV-LEDs, possibly resulting from the additional series resistance between chips. Figure 2(c) shows the forward voltage as the function of the current density. The forward voltage initially increased and then decreased upon increasing the injection current density. The maximum voltage occurred at current density 197 and 205 A/cm2 for the ST and HV UV-LEDs, respectively. It was found that there existed the kinks for the both J-V curves. The kinks of J-V curves resulted from the high current injection which resulted in the junction temperature increasing and further reduced the bandgap of LEDs. Although there existed abnormal J-V curves for both DUV LEDs, the effect on the voltage decreasing was much serious for the ST DUV LEDs than that for the HV DUV LEDs. It was attributed to the fact that the operation current of ST DUV LEDs was higher than that of HV DUV LEDs. Note worthily, the efficiency of DUV-LEDs was not good enough. The junction temperature could be generated in DUV-LEDs as the current density, further increasing and affecting the energy band and reducing the forward voltage. Compared with the voltage variation of the two samples, the voltage of the ST sample decreased more seriously than that of the HV sample. ST decreased by about 0.6 V at 280 A/cm2, and HV decreased by about 0.3 V. This phenomenon may be due to the current crowding effect and overflowing into the cladding layer in the ST sample as injection high current, allowing the temperature to affect significantly more than the HV sample. In addition, the loading capability in the input current density of HV DUV-LED was also higher than that of the traditional ST sample.
Fig. 2. (a) Current and dynamic resistance as a function of voltage, forward voltage as a function of (b) injection current and (c) injection current density.
The emission output powers as the function of injection currents for ST and HV DUV-LEDs were measured, shown in Fig. 3(a). The HV sample obtained a higher output power. When the injection current was 320 mA, the maximum output power was about 47.1 mW in the HV sample. At the same injection current, the ST sample only presented 32.38 mW output power. Nevertheless, the highest output power of the ST sample was 43.1 mW at 480 mA input current, which is also lower than the HV performance. Given that the injection area of the p-region of HV only had half of ST and that the injection current was the same, the current density in HV DUV-LED was higher than that of the ST DUV-LED. Hence, the output power of the HV sample was higher than that of the ST sample. On the other hand, it also resulted in degradation occurring at the early current injection. For a fair comparison, Fig. 3(b) presents the corresponding output power as the function of the injection current density. Notably, the two samples almost had similar output power in the small current density, from 0.004 A/cm2 to 100 A/cm2. ST had a slightly higher output power in the 100 A/cm2 to 200 A/cm2 region than the HV sample due to a smaller resistance characteristic. The resistance is listed in Table 1. However, as the current density further increased, ST decayed obviously from 43.1 mW to 34.9 mW. Meanwhile, HV further improved the output power to 47.1 mW, slowly dropping off as the injection current density increased from 280 A/cm2 to 600 A/cm2. Therefore, considering the input current density, the HV DUV-LEDs also performed better than the ST at the higher current density because half of the active area offered a better current spreading condition in the HV sample. Furthermore, the current crowding effect would affect the properties in ST DUV-LEDs, further decreasing the performance of the electric-optic characteristics. In addition to the current spreading, the mirror deposited on the p and n pads shown in Fig. 1 (i) and the inset of Fig. 3(b) in the HV sample could also enhance the optical emission output power. The lit-up pictures for these two samples were observed in Fig. 3(c). Obviously, with the same input current density at 8 A/cm2 and 40 A/cm2, the emission area and intensity distribution in the HV sample were larger and more uniform than those in the ST sample. The emission areas for the HV and ST LEDs were 84975 μm2 and 81450 μm2, respectively. The finger electrode limited the current spreading, and the current crowding effect occurred in higher current density injected, which reduced the output power. Meanwhile, the maximum of WPE in these two samples was calculated and listed in Table 1. The HV sample had a higher electric-optic conversion efficiency than the ST sample. The values of the maximum WPE were 2.46% and 2.28% in HV and ST DUV-LEDs, respectively. On the other hand, the brightest area occurring in the mirror edges shown in the insets of Fig. 3(a) and (b) not only indicated that the mirrors played their reflective function but also indicated that the mirror area in the HV DUV-LED was larger than that of ST DUV-LED.
Fig. 3. Emission output power as the function of (a) injection current, (b) injection current density, and (c) emission images in the same current density. The insets shown in Fig. (a) and (b) were the top and backside view images of ST and HV DUVLEDs, respectively, measured by an optical microscope.
Table 1. Chips size and opto-electronic characteristics of the ST and HV DUV-LEDs in this study
The value of EQE as the function of the input current after measurement and calculation is shown in Fig. 4(a). When the injection current was below 420 mA, the EQE values of HV were better than those in the ST sample. In 450 mA to 500 mA injection current, the droop of output power makes the EQE value of the HV sample lower than that of the ST sample. Table 1 also lists the maximum EQE values of HV and ST—that is, 5% and 2.5%, respectively. Because the operation voltages were different for these two DUV LEDs, efficiency droop could be considered as function of input electrical power and shown in Fig. 4(b). The EQE value of HV was twice as higher as ST when the input power increased from 0.1 W to 5 W. Not only, the EQE of HV LEDs drooped from 5.01% to 3.33% corresponding the input power was 0.4 W to 5.1 W. Correspondingly, the EQE of ST LEDs drooped from 2.53% to 1.01% corresponding the input power was 0.34 W to 5 W. The EQE droop rate was 33.5 and 59.7% for the HV and ST DUV LEDs, respectively. Furthermore, the EQE value of the ST sample had a significant decrease due to the output power decay in the input power from 3.5 W to 4 W. This phenomenon may have resulted from the temperature affecting the ST sample in higher input power, narrowing the bandgap and decreasing the output power. The temperature effect can be evaluated by wavelength measurement. As the heat of LEDs increases, the emission wavelength will shift toward a long wavelength. Figure 4(b) presents the peak wavelength as the input power increased from 0.1 to 5 W. It can be observed that the obvious red shift in the ST sample increased from 278 nm to 282 nm and only shifted from 278.5 nm to 280.5 nm for the HV LEDs in the same input power. The higher position of the HV sample in the initial input power was due to the series resistance, generating an extra temperature in the HV sample. In addition, high voltage and low current characteristics resulted in higher resistance in the HV DUV-LED, leading to a higher wavelength in the 0.1 W input power. At the same electrical power, the injection current of ST DUV LEDs was higher than HV DUV LEDs. The high current injection resulted in the current crowding and overflowing into the cladding layer. These make a larger wavelength shift and EQE dope in the high electrical power region for the ST DUV LEDs. Nevertheless, the current crowding in the ST sample made the heating effect generation in DUV-LEDs occur more frequently than in the HV sample.
Fig. 4. EQE as the function of (a) injection current, (b) input power (left axis), and peak wavelength as the function of input power (right axis).
Finally, the view angles of HV and ST DUV LEDs were measured, as shown in Fig. 5. In the same current density shown in Fig. 5(a), the view angles were almost the same in these two samples. However, the intensity of the HV sample was higher than that of ST in the range of 30 to 45 degrees due to the emission pattern in the HV sample being more uniform. Two series chips also contributed to more light escaping from the side walls. In addition, the traditional finger electrodes limited the current spreading, further affecting the emission pattern. Furthermore, the obtained results were consistent with the emission images shown in Fig. 3(c). The results of the view angles when the input current was both at 100 mA in the two samples are presented in Fig. 5(b). It can be observed that the view angle of the HV sample had a wider angle than ST; the view angles were 130 degrees and 90 degrees, respectively. Its intensity was also twice higher than the ST sample due to a higher current density in the HV sample.
Fig. 5. View angles of (a) the same injection current density and (b) the same injection current of ST and HV DUV-LEDs.
4. Conclusion
By connecting with two series contact DUV-LEDs, the characteristic of HV DUV-LEDs operating at high-voltage and low-current was obtained, which could be directly applied to consumer products without a transformer’s components in the circuit. HV showed higher dynamic resistances than ST—5.9 and 2.5 Ω at the injection current density of 120 A/cm2, respectively. The maximum output power was exhibited at 47.1 mW in the HV sample at an injection current of 320 mA, while the ST sample only presented 32.38 mW output power. Further, the HV sample had a higher electric-optic conversion efficiency than ST; the values of the maximum WPE were 2.46 and 2.28% in HV and ST DUV-LEDs, respectively. HV also exhibited a better EQE value than ST—the maximum value of HV was twice higher than ST due to the excellent current spreading and mirror effect, which enhanced the characteristics of emission output power and EQE. The maximum EQE values of HV and ST were 5.48% and 2.57%. As the input power increased, there were more redshifts in the ST sample (from 278 to 282 nm). On the other hand, the HV sample only had redshifts of 278.5 to 208.5 nm in the same input power. While the view angles were almost the same in these two samples at the same injection current density, the intensity of the HV sample was higher than ST in the range of 30 to 45 degrees. In addition, the traditional finger electrodes limited the current spreading, further affecting the emission pattern. When the input current was both 100 mA in the two samples, the view angle of the HV sample had a wider view angle than ST—the view angles were 130 and 90 degrees, respectively. Its intensity was also twice higher than that in ST due to higher current density in the HV sample
Funding
Ministry of Science and Technology, Taiwan (MOST 110-2224-E-A49-003, 111-2218-E-A49-019-MBK).
Acknowledgments
This study was also supported by Science Park Emerging Technology Application Program, Taiwan, R.O.C., under the grants 111AO29B. We acknowledge MA Tek for funding and material measurement support (grant number: 2022-T-018). We thank the National Nano Device Laboratory and Taiwan Instrument Research Institute for allowing us to use their facilities.
Disclosures
The authors declare no conflicts of interest.
Data availability
The data underlying the results presented in this paper are not publicly available presently but may be obtained from the authors upon reasonable request.
References
1. H. Amano, H. Amano, R. Collazo, C. De Santi, S. Einfeldt, M. Funato, J. Glaab, S. Hagedorn, A. Hirano, H. Hirayama, R. Ishii, Y. Kashima, Y. Kawakami, R. Kirste, M. Kneissl, R. Martin, F. Mehnke, M. Meneghini, A. Ougazzaden, P. J. Parbrook, S. Rajan, P. Reddy, F. Römer, J. Ruschel, B. Sarkar, F. Scholz, L. J. Schowalter, P. Shields, Z. Sitar, L. Sulmoni, T. Wang, T. Wernicke, M. Weyers, B. Witzigmann, Y.-R. Wu, T. Wunderer, and Y. Zhang, “The 2020 UV emitter roadmap.,” J. Phys. D: Appl. Phys. 53(50), 503001 (2020). [CrossRef]
2. J. Lu, Y. Zhong, and S. Zhao, “Intrinsic excitation-dependent room-temperature internal quantum efficiency of AlGaN nanowires with varying Al contents,” J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom. 39(2), 022803 (2021). [CrossRef]
3. C. Huang, H. Zhang, and H. Sun, “Ultraviolet optoelectronic devices based on AlGaN-SiC platform: Towards monolithic photonics integration system,” Nano Energy 77, 105149 (2020). [CrossRef]
4. M. A. Khan, N. Maeda, J. Yun, M. Jo, Y. Yamada, and H. Hirayama, “Achieving 9.6% efficiency in 304 nm p-AlGaN UVB LED via increasing the holes injection and light reflectance.,” Sci. Rep. 12(1), 2591 (2022). [CrossRef]
5. Y. T. Liao, C. Thomidis, C. K. Kao, and T. D. Moustakas, “AlGaN based deep ultraviolet light emitting diodes with high internal quantum efficiency grown by molecular beam epitaxy,” Appl. Phys. Lett. 98(8), 081110 (2011). [CrossRef]
6. M. Crawford, “Materials challenges of AlGaN-based UV optoelectronic devices,” Semicond. Semimetals 96, 3–44 (2017). [CrossRef]
7. V. K. Sharma, S. T. Tan, H. Y. Zheng, S. Shendre, A. Baum, F. Chalvet, J. Tirén, and H. V. Demir, “On-Chip Mercury-Free Deep-UV Light-Emitting Sources with Ultrahigh Germicidal Efficiency,” Adv. Opt. Mater. 9(15), 2100072 (2021). [CrossRef]
8. M. S. Shur and R. Gaska, “Deep-ultraviolet light-emitting diodes,” IEEE Trans. Electron Devices 57(1), 12–25 (2010). [CrossRef]
9. C. Chu, K. Tian, Y. Zhang, W. Bi, and Z. H. Zhang, “Progress in external quantum efficiency for III-Nitride based deep ultraviolet light-emitting diodes,” Phys. Status Solidi A 216(4), 1800815 (2019). [CrossRef]
10. T. C. Hsu, Y. T. Teng, Y. W. Yeh, X. Fan, K. H. Chu, S. H. Lin, K. K. Yeh, P. T. Lee, Y. Lin, Z. Chen, T. Wu, and H. C. Kuo, “Perspectives on UVC LED: Its progress and application,” Photonics 8(6), 196 (2021). [CrossRef]
11. Y. Matsukura, T. Inazu, C. Pernot, N. Shibata, M. Kushimoto, M. Deki, Y. Honda, and H. Amano, “Improving light output power of AlGaN-based deep-ultraviolet light-emitting diodes by optimizing the optical thickness of p-layers,” Appl. Phys. Express 14(8), 084004 (2021). [CrossRef]
12. S. H. Lin, M. C. Tseng, K. W. Peng, S. Lai, M. C. Shen, R. H. Horng, S. Y. Lien, D. S. Wuu, H. C. Kuo, T. Wu, and Z. Chen, “Enhanced external quantum efficiencies of AlGaN-based deep-UV LEDs using reflective passivation layer,” Opt. Express 29(23), 37835–37844 (2021). [CrossRef]
13. C. Y. Kang, C. H. Lin, T. Wu, P. T. Lee, Z. Chen, and H. C. Kuo, “A Novel Liquid Packaging Structure of Deep-Ultraviolet Light-Emitting Diodes to Enhance the Light-Extraction Efficiency,” Crystals 9(4), 203 (2019). [CrossRef]
14. W. H. Zhang, W. K. Lin, C. T. Yeh, S. B. Chiang, and C. S. Jao, “A novel liquid-packaging technology for highly reliable UV-LED encapsulation,” Heat Transfer. Res. 50(4), 349–360 (2019). [CrossRef]
15. M. Ichikawa, A. Fujioka, T. Kosugi, S. Endo, H. Sagawa, H. Tamaki, T. Mukai, M. Uomoto, and T. Shimatsu, “High-output-power deep ultraviolet light-emitting diode assembly using direct bonding,” Appl. Phys. Express 9(7), 072101 (2016). [CrossRef]
16. S. H. Liang and W. H. Sun, “Recent Advances in Packaging Technologies of AlGaN-Based Deep Ultraviolet Light-Emitting Diodes.,” Adv. Mater. Technol. 7, 2101502 (2022). [CrossRef]
