GaN-based Near-Ultraviolet Light-Emitting Diodes (Near-UV LEDs)

Pioneering work of Pankove, Akasaki, Amano, Schubert, and Nakamura has led to the development of bright green and blue LEDs based on nitride semiconductors in the recent decade. The first blue light-emitting diode (LED) using III-nitrides materials was fabricated by J. I. Pankove et al. [15] in 1972. Since that, related research about GaN-based material is going on continually. However, progresses have been limited because of highly background n-type concentration resulting from the native defects commonly thought to be nitrogen

vacancies and residual impurities such as silicon and oxygen acted as an efficient donor, poorly conducting p-type GaN, and the lack of appropriate substrates for epitaxial growth.

Until late 1980s, I. Akasaki and H. Amano et al. discovered a very useful method of an AlN buffer layer and developed low-energy electron beam interaction (LEEBI) techniques to obtain a better quality GaN film and conductive p-type GaN layer [16-18]. The quality of GaN thin film grown by atmospheric pressure metal-organic vapor phase epitaxial (AP-MOVPE) using AlN buffer layers is shown to be excellent in terms of morphological, crystalline, and optical properties. Using AlN buffer layers, GaN thin films with optically flat surfaces free from cracks are successfully grown. The narrower x‐ray rocking curve of GaN film with AlN buffer layer from the (0006) plane is shown in Fig. 1.2.1 [16]. In 1989, I.

Akasaki et al. put a sample of GaN:Mg in a scanning electron microscope (SEM) with an optical window through which they could see the blue cathodoluminescence. As shown in Fig.

1.2.1, they noticed that the intensity of luminescence kept getting stronger the longer they scanned. When the brightness seemed to saturate they took the sample out and measured the Hall effect. To their great surprise and to the world’s astonishment, the previously insulating sample had become conducting p-type [17].

Fig. 1.2.1. X-ray rocking curve for (0006) diffraction from GaN grown at 970 °C with the AlN buffer layer. Dotted line shows data obtained by HVPE.

Fig. 1.2.2. Photoluminescence spectra of GaN:Mg (b) before and (a) after the electron beam treatment. The ratio of the two peaks is 100.

Until 1991, S. Nakamura developed a new buffer layer for high quality GaN growth by MOVPE. Strenuous efforts were made to optimize growth conditions by introducing suitable

buffer layers (between the sapphire substrate and the active GaN) of GaN grown at low temperatures (~ 500 °C) and by adjusting flow rates and temperatures with the object of reducing n-type background doping levels and increasing Hall mobility. The optimum thickness of the GaN buffer layer was around 200 Å . Fig. 1.2.3 shows the Hall mobility and carrier concentration measured at 77 K and 300 K as a function of the thickness of the GaN buffer layer.This eventually proved successful high Hall mobility values of 600 cm²/V．s at 300 K and background levels below 10¹⁷ cm^-3 were achieved, followed by the important breakthrough of p-type doping at the end of the 1980s [19].

Fig. 1.2.3. The Hall mobility and carrier concentration measured at 77 K and 300 K as a function of the thickness of the GaN buffer layer.

These discoveries initiated a new strong interest in this research field. Finally, the first GaN-based blue LED constructed of a real p-n junction was achieved, which had greatly improved in the device performance. However, the acceptor concentration of p-type GaN is still too low such that the application of these materials is still unreliable. After that, in 1992, S. Nakamura et al. made the definitive experiment that showed by thermal annealing in nitrogen ambient that above 700°C the passivating H was dissociated from Mg, rendering

temperatures. H could be reintroduced by heating in NH₃ making GaN:Mg insulating again [20]. Afterward Nakamura and Mukai et al. [21] succeeded in growing high-quality InGaN films that emitted strong band-to-band emission from green to UV region by changing the indium content of InGaN with a two-flow MOCVD method. Nowadays, most of III-Nitride based LEDs use InGaN as active layer instead of GaN, since adding a small amount of indium into the GaN is very important to obtain a strong band-to-band emission at RT.

Fig. 1.2.4. Resistivity of GaN:Mg after annealing at various temperatures.

The most spectacular application and the first commercial GaN product is the LED. In 1997, Nakamura et al. have produced the brightest LEDs using GaN p-n junctions that include a Zn-doped well as shown in Fig. 1.2.5 [22]. The purpose of the well is to confine the carriers in a small volume, and the purpose of the Zn is to introduce an efficient blue or green luminescent center [23].

Further development of the AlN, AlGaN and AlInGaN materials resulted in an appearance of ultraviolet (UV) LEDs which represent the next frontier in solid-state optoelectronics with a large potential in biological, medical and environmental instrumentation, resin curing, photo-catalyst for disinfection and deodorization, UV light source [24], and there have been interests in solid-state lighting by using near-UV LEDs light

for the phosphor-converting source [25, 26].

Fig. 1.2.5. Structure of III-Nitride light emitting diodes (LEDs).

Near-UV LED has some unique advantages, such as safer operation with very low heat generation, low electricity consumption, without UVB and UVC dangerous wavelengths, no warm-up time, instant turn on/off, lower operating voltage, compact and scalable design for ease of integration, much longer service life, and environmentally friendly with no ozone emissions or dangerous mercury. Fig. 1.2.6 shows extensively applications of Near-UV LEDs.

Fig. 1.2.6. Applications of Near-UV LEDs.

A great progress in solid-state sources of GaN-based p-n junction UV light was achieved by Akasaki et al. in 1992 [4]. Present UV LEDs are based on heterostructures developed

[29], and quaternary AlInGaN [30, 31]. There are two main issues of GaN-based UV LED:

the chips must feature band diagrams structure that assists high efficiency of carrier injection into the active layer, the internal quantum efficiency (IQE) should be maximized by enhancing radiative recombination and restraining the nonradiative recombination, and light generated in active layer must be efficiently extracted from the chip. Certainly, reduction of the dislocation density, solving self-absorption of GaN, and preventing cracking of epitaxial layers mismatched to the substrate is one of the most important issues.

Most of GaN-based UV LEDs are grown on sapphire substrate, which has a 16 % lattice mismatch with GaN. This disadvantage is being bypassed through dislocation reducing by epitaxial lateral overgrowth (ELO) and by using superlattices, strain-compensating layers and quaternary AlInGaN alloys. To increase the IQE of UV LEDs, optimization of quantum well structures is necessary through selecting composition and doping profiles of the well, barrier and p-type layers, shaping of the interfaces between well and barrier, and engineering of the built-in electric field to avoid the quantum-confined Stark effect (QCSE). In addition, basic investigation for unveiling the routes of nonradiative recombination in AlGaN alloys with high molar fraction of aluminum is needed. To overcome the high resistivity of p-AlGaN, novel doping approaches including piezoelectric and super-lattice doping, graded composition, and co-doping of magnesium (Mg) and indium (In) are being developed.

First commercial 375 nm LEDs were fabricated by Nichia in 1998 [28]. Typically, these devices feature 1.5~2 mW light output power and are available with the out-coupling optics for narrow-angle (20 °) and wide-angle (110 °) far-field radiation pattern. Cree fabricated the first near-UV LED for use in the illumination market in 2001. The InGaN based devices were grown on SiC substrate, and the light output power was about 12 mW with the wavelength between 395~405 nm. These near-UV LEDs have a geometrically enhanced vertical chip structure to maximize light extraction efficiency and require only a single Au-wire bond

connection with the device.

Shown in Fig. 1.2.7 is the benchmark of history development of GaN-based UVA LED in the wavelength range from 320 to 410 nm [32]. From the benchmark, it can be deliberated that the UV LED has an energetic development after 1998.

Fig. 1.2.7. Benchmark of UVA LED in the wavelength range from 320 to 410 nm.

Although the excellent external quantum efficiency (EQE) that have been obtained by many groups for LEDs in the near UV range, the EQE of InGaN-based UV LEDs with emission wavelength shorter than 365 nm is at least one order of magnitude below the best devices in the near-UV wavelength range about 400 nm. There are multiple causes that currently limit the EQE of GaN-based UV LEDs. For example, with decreasing wavelengths, the light output power is dropping and challenges in growing high quality nitride heterostructures with a high aluminum (Al) molar fraction are becoming more difficult [30].

The solutions to device problems lie in using better substrates (ex: AlN substrates in non-polar orientations, and patterned sapphire substrate), using better epitaxial growth techniques (ex: ELO), improving device design (ex: vertical chip and flip chip) and using better contact technology.