According to current market reflections, LEDs have gradually replaced traditional light sources, and are widely used in display screens, traffic control, outdoor lighting and other fields. In order to enable more consumers to understand LED, the following Qipu Optoelectronics will give you some tips from the technology of LED materials.
1. Epitaxy technology
Metal organic chemical vapor deposition (MOCVD) technology is the mainstream technology for growing LEDs. In recent years, thanks to the advancement of MOCVD equipment, the cost of LED material epitaxy has dropped significantly. The main equipment providers on the market are Aixtron of Germany and Veeco of the United States.
The former can provide two types of equipment: a horizontal planetary reaction chamber and a near-coupled shower head reaction chamber. Its advantages are that it saves raw materials and grows LED epitaxial wafers with good uniformity. The latter equipment uses the high-speed rotation of the tray to generate laminar flow, which has the advantages of simple maintenance and large capacity.
In addition, Nippon Acid produces atmospheric MOCVD exclusively for Japanese companies to obtain better crystal quality. The American Applied Materials Corporation originally created the multi-chamber MOCVD equipment, and has already begun trial use in the industry.
The future development direction of MOCVD equipment includes: further expanding the volume of the reaction chamber to increase production capacity, further improving the utilization of MO source, ammonia and other raw materials, further improving the on-site monitoring capability of the epitaxial wafer, and further optimizing the temperature field and air flow field Control to improve the ability to support epitaxy of large-size substrates.
2. Substrate
(1) Graphic substrate
The substrate is the base supporting the epitaxial film. Due to the lack of a homogeneous substrate, GaN-based LEDs are generally grown on heterogeneous substrates such as sapphire, SiC, and Si. So far, sapphire has become the most cost-effective substrate and is the most widely used. As the refractive index of GaN is higher than that of sapphire, in order to reduce the total emission of the light emitted from the LED at the interface of the substrate, the current front-mounted chip generally conducts material epitaxy on the graphic substrate to improve the light scattering.
Common graphic substrate patterns are generally close-packed hexagonal cone arrays with a size on the order of micrometers, which can increase the light extraction efficiency of the LED to more than 60%. At the same time, studies have shown that the use of a patterned substrate combined with a certain growth process can control the extension direction of dislocations in GaN, thereby effectively reducing the dislocation density of the GaN epitaxial layer. For a considerable period of time in the future, the graphics substrate will still be the main technical means adopted by the formal chip.
In the future, the development direction of graphic substrates is toward smaller sizes. At present, limited by the manufacturing cost, the sapphire pattern substrate is generally manufactured by contact exposure and ICP dry etching, and the size can only be on the order of micrometers. If the size can be further reduced to the order of hundreds of nanometers comparable to the wavelength of light, the light scattering ability can be further improved. It can even be made into a periodic structure, and the physical effect of the two-dimensional photonic crystal can be used to further improve the light extraction efficiency. The manufacturing methods of nano-patterns include electron beam exposure, nano-imprinting, and self-assembly of nano-spheres. In terms of cost, the latter two are more suitable for substrate processing.
(2) Large size substrate
At present, 2-inch sapphire substrates are still the mainstream in the industry. Some large international manufacturers are already using 3-inch or even 4-inch substrates, and it is expected to expand to 6-inch substrates in the future. The expansion of the size of the substrate is conducive to reducing the edge effect of the epitaxial wafer and improving the yield of LEDs. However, the current large-size sapphire substrate is still expensive, and the supporting material epitaxy equipment and chip process equipment must be upgraded after the substrate size is expanded, which is a significant investment for manufacturers.
(3) SiC substrate
The lattice mismatch between the SiC substrate and the GaN-based material is even smaller. Facts have proved that the quality of the GaN crystal grown on SiC is slightly better than the result on the sapphire substrate. However, the manufacturing cost of SiC substrates, especially high-quality SiC substrates, is very high, so few manufacturers use LED materials for epitaxy. However, with its own manufacturing advantages on high-quality SiC substrates, Cree of the United States has become the only manufacturer in the industry that only grows LEDs on SiC substrates, thus avoiding the patent barrier of growing GaN on sapphire substrates. At present, the mainstream size of SiC substrate is 3 inches, and it is expected to expand to 4 inches in the future. Compared with sapphire substrates, SiC substrates are more suitable for making GaN-based electronic devices. With the development of wide-bandgap semiconductor power electronic devices in the future, the cost of SiC substrates is expected to be further reduced.
(4) Si substrate
The Si substrate is regarded as an ideal choice for reducing the cost of LED epitaxial wafers, because its large-size (8-inch, 12-inch) substrates are the most mature. However, because the lattice mismatch and thermal mismatch are too large and difficult to control, the quality of the LED material based on the Si substrate is relatively poor, and the yield is low, so LED products based on the Si substrate on the market are very rare. At present, LEDs grown on Si mainly use substrates below 6 inches. Considering the yield factor, the actual cost of LEDs is not superior to those based on sapphire substrates. Like SiC substrates, most research institutions and manufacturers prefer to grow electronic devices on Si substrates instead of LEDs. In the future, LED epitaxial technology on Si substrates should target larger substrates such as 8 inches or 12 inches.
(5) Homogeneous substrate
As mentioned earlier, the current epitaxial growth of LEDs is still dominated by the epitaxial growth of heterogeneous substrates. However, the homogeneous substrate with lattice matching and thermal matching is still regarded as the ultimate solution to improve the crystal quality and LED performance. In recent years, with the development of hydride vapor deposition (HVPE) epitaxial technology, the manufacturing technology of large-area GaN-based thick substrates has received attention. The manufacturing method is generally to use HVPE to rapidly grow on heterogeneous substrates to obtain tens to The GaN bulk material with a thickness of hundreds of microns is then peeled off the thick GaN film from the substrate by mechanical, chemical or physical means, and the GaN thick layer is used as the substrate for LED epitaxy.
3. Epitaxial structure and epitaxial technology
(1) Droop effect
After several years of development, the epitaxial layer structure and epitaxial technology of LED have been relatively mature, and its internal quantum efficiency can reach more than 90%. However, in recent years, with the rise of high-power LED chips, the quantum efficiency drop of LEDs under large injection has attracted widespread attention. This phenomenon is vividly called the Droop effect. As far as the industry is concerned, solving the Droop effect can further reduce the chip size under the premise of ensuring power, and achieve the purpose of reducing costs. For academia, the cause of the Droop effect is a hotspot that attracts scientists to study.
Unlike traditional semiconductor optoelectronic materials, the cause of the Droop effect of GaN-based LEDs is very complicated, and there is a corresponding lack of effective solutions. Researchers have explored and found that several reasons for their preference are: delocalization of carriers, leakage or overflow of carriers from the active region, and Auger recombination. Although the specific reason is not clear, experiments have found that using a wider quantum well to reduce the carrier density and optimizing the electron blocking layer in the p-type region are both means to alleviate the Droop effect.
(2) Quantum well active area
The active region of the InGaN/GaN quantum well is the core of the LED epitaxial material. The key to growing the InGaN quantum well is to control the stress of the quantum well and reduce the influence of the polarization effect. Conventional growth techniques include: growing a low-In composition InGaN pre-well before multiple quantum wells to release stress and act as a carrier reservoir, growing a GaN barrier layer at elevated temperature to improve the crystal quality of the barrier layer, and growing a lattice-matched InGaAlN barrier layer Or growth stress complementary InGaN/AlGaN structure, etc. There is no uniform standard for the number of quantum wells. The number of quantum wells used in the industry ranges from 5 to 15, and the final effect is not much different. LEDs with a smaller number of wells have higher efficiency under small injections, but a larger number of wells LEDs are more efficient under large injections.
(3) p-type region
The p-type doping of GaN is one of the important bottlenecks that plagued the production of LEDs in the early stage. This is because the unintentionally doped GaN is n-type, and the electron concentration is above 1×1016 cm-3, which makes it difficult to realize p-type GaN. The most successful p-type dopant so far is Mg, but it still faces problems such as lattice damage caused by high-concentration doping, and the acceptor is easily passivated by the H element in the reaction chamber. The oxygen thermal annealing method invented by Shuji Nakamura at Nichia is simple and effective. It is a widely used acceptor activation method. Some manufacturers directly use nitrogen in-situ annealing activation in the MOCVD epitaxial furnace. Nichia's p-GaN quality is the best, which may be related to the atmospheric pressure MOCVD growth process.
In addition, there are some reports on the use of p-AlGaN/GaN superlattices and p-InGaN/GaN superlattices to increase the hole concentration. Nevertheless, the hole concentration and hole mobility of p-GaN are still very different from the electrons of n-GaN, which causes the asymmetry of LED carrier injection. Generally, an electron blocking layer of p-AlGaN must be inserted on the side of the quantum well close to the p-GaN. However, the polarity mismatch between AlGaN and the quantum well region is considered to be the main cause of carrier leakage, so some manufacturers have recently tried to use p-InGaAlN instead.
4. No phosphor single chip white light LED
The existing white light LED mainly uses a combination of blue LED and yellow phosphor to emit white light. The typical color rendering index of this white light is not high, especially the reproducibility of red and green is weak. In addition, phosphors also face problems such as poor reliability and loss of efficiency. It is theoretically feasible to rely entirely on InGaN material as the light-emitting area to achieve white light in a single chip.
5. Other color LED
The external quantum efficiency of GaN-based blue LEDs has exceeded 60%, which means that blue LED devices have been relatively mature. Therefore, people began to focus on other wavebands that can be covered by nitride materials. Traditional III-V semiconductors have been very mature in producing infrared and red light-emitting devices, so it is more meaningful for nitride to develop green and ultraviolet LEDs.
(1) Green LED
The green light band is currently the least efficient in the visible light band and is called "GreenGap". The reason for the low efficiency of InGaN in the green band is that the polarization effect caused by the higher In composition and the wider quantum well becomes stronger. The aforementioned growth of non-polar/semi-polar surface LEDs is an effective method to improve the efficiency of green LEDs, but it is currently not practical due to the limitation of homogeneous substrates.
(2) UV LED
Ultraviolet light has important applications in fields such as curing, sterilization, early warning, and covert communications. Traditional ultraviolet light sources are all vacuum devices. Nitride materials are the most suitable material for making UV LEDs. However, due to the high dislocation density and the light-emitting area of AlGaN (does not contain In, the advantage of InGaN luminous efficiency is not sensitive to dislocations), GaN-based UV LEDs are especially The efficiency of deep ultraviolet LED (wavelength below 280nm) is still very low. The Riken Institute in Japan and the ArifKhan group at the University of Southern California in the United States are pioneers in the study of deep-ultraviolet LEDs. Riken can achieve the external quantum efficiency of deep ultraviolet LEDs to 3.8%, and the output power can reach 30mW.
