One of the characteristics of LEDs is their low-current, low-voltage-drive power-saving features. Such features attract the attention of everyone when the world's energy resources are lacking and countries are promoting green concepts. In addition to efforts to develop new energy sources, governments around the world have also put considerable effort into improving the effectiveness of existing electrical equipment and environmental protection. While R&D is about how to reduce industrial electricity consumption, the power consumption of household appliances, which currently have a penetration rate of about 80%, is gradually being taken seriously. In terms of lighting, if we replace the conventional 60W incandescent bulbs with fluorescent lamps (66-75lm/W) with higher luminous efficacy, the annual lighting time of 3500 hours will save approximately 6.89 yuan per year. Billion degrees (about 88,600 kW).
Although fluorescent lamps have higher luminous efficacy and lower manufacturing cost, fluorescent lamps contain mercury, and the materials used for encapsulating fluorescent lamps are mainly ultraviolet-absorbing glass. The fragile nature of glass, coupled with the non-recoverability of mercury waste, can cause serious environmental pollution. Therefore, the EU has explicitly ordered that these mercury-containing products be banned in 2007. As a result, the development of new lighting sources has become the goal of the governments of various countries. LED (light emitting diode), which is what we usually refer to as light-emitting diodes, is more Lighting development focus.
Light-emitting diode light-emitting diode The so-called light-emitting diode is basically a conventional pn diode, but its main function is not to rectify the current, but to use its positive current to pass through the pn junction to promote the junction. Some of the electron holes combine and emit light, and their light emission characteristics can be referred to FIG. 1 .
The wavelength of the light emitted by the light emitting diode, in addition to the wavelength of the semiconductor material used for the diode, also depends on the mixing ratio between different materials. Fig. 2 shows the relationship between energy band, lattice constant, and emission wavelength of each luminescent material. It can be seen that red, yellow and green light are mainly based on InGaAlP materials, while blue and green light are mainly based on InGaN materials.
The technology of light-emitting diodes For semiconductor light-emitting diodes, the matching of the lattice is a major issue, because for most III-V semiconductors, there is not just a suitable substrate to carry the epitaxial layer above. The growth of the epitaxial layer must match the crystal lattice size of the substrate, so as not to cause lattice defects due to the force-related factors, so that the photons emitted by the device are absorbed by defects, and the luminous efficacy of the device is greatly reduced. The earliest heterogeneity of III-V semiconductor epitaxy (heteroepitaxy) is to use GaAs as a substrate and generate an epitaxial layer of GaAlAs on it, because the lattices of these two materials are very similar, so the epitaxial layer and the substrate The stress between the two is extremely small, so there is not much trouble in the R&D process. However, there is a problem with the stress that epitaxial growth such as GaAs1-xPx grows on a GaAs substrate or GaAsxP1-x grows on a GaP substrate. Therefore, in photoelectric materials, the ratio of binary, ternary, or even quaternary materials is often adjusted so that one side can match the lattice structure of the substrate in addition to the proportions of the plural atoms of different sizes, and can also adjust the semiconductors. The energy gap size is used to adjust the wavelength of light emitted by the light emitting device. This method is also complicated in the adjustment of the epitaxial parameters. Therefore, it can be seen that the epitaxial technology can be said to be the core of the semiconductor light emitting device technology.
While the epitaxial method is being promoted, the epitaxial structure is also continuously improving. The earliest structures were of course conventional pn-junction light-emitting diodes, but their luminous efficacy could not be significantly improved. Therefore, a method using a single heterojunction (SH) structure began to be used in an epitaxial process. The improvement of the minority carrier injection efficiency in the diodes results in a significant increase in luminous efficacy. Later, the Double Heterojunction (DH) structure was developed. The gap between the two sides of the structure is higher than that of the middle one. Therefore, it is very effective to inject bilateral carriers into the intermediate layer and completely trap the carriers. Within this range, very high photoelectric conversion efficiency is generated. The latest method is of course to use a quantum structure in the epitaxial layer. When the thickness of the intermediate layer of the double heterojunction structure is gradually reduced to several tens of angstroms (A), electrons or holes generate quantum effects, which can significantly increase the photoelectricity. The effect of the conversion.
The epitaxial technology mentioned here is mainly aimed at the GaAs series of materials in the III-V group where the emission wavelength is concentrated in the red and yellow wavelength bands. This series of light-emitting diodes developed earlier and also achieved better results earlier. However, if it is desired to obtain a full-color semiconductor light source, semiconductor light-emitting diodes in blue and green wavelength bands must be developed in any case, and GaN-based light-emitting diodes also have such requirements, and there have been noticeable improvements in recent years. In particular, the development history and evolution of GaN materials and their current development are summarized as follows.
The development of GaN materials - blue light-emitting diodes used in blue, green light-emitting diode materials, early mainly ZnSe and GaN. Because of the reliability of ZnSe, GaN has more room for development. However, the early progress of GaN research has not been able to achieve significant progress. The main reason is that it has not been able to find substrates that match the GaN lattice constants, resulting in an excessively high degree of defect integration in the epitaxial crystals. Therefore, the luminous efficacy cannot always be improved. Another reason for the failure of GaN to achieve a breakthrough is that the P-GaN portion of the device is not easily generated. Not only do the P-GaN doping is too low, but the mobility of the holes is also lower. In this way, it was not until 1983 that Japan's Yoshida (S. Yoshida) and others used high-temperature aluminum nitride (AlN) as a buffer layer on a Sapphire substrate, and then the resulting GaN was better. After crystallization, Professor Akazaki and others of Nagoya University used MOCVD to grow the AlN buffer layer at a low temperature (600oC) to obtain a mirror-like GaN grown at a high temperature. In 1991, Nichiamura, a researcher of Nichia Co., used low-temperature growth of an amorphous buffer layer of GaN, and then grew at high temperature to obtain mirror-like GaN. At this time, the problem of epitaxial growth has been significant. Breakthrough. On the other hand, in 1989 Prof. Akazyong used electron beam irradiation of magnesium (Mg)-doped P-GaN to obtain apparent P-type GaN, and Nisei's Nakamura's thermal annealing at 700°C was then used to directly complete P-type GaN. The manufacture of GaN has finally led to breakthroughs in the two major problems that have plagued GaN.
In 1993, Nichia used the above two studies to successfully develop a GaN blue light emitting diode that emits a candlelight (Candela) with a lifetime of tens of thousands of hours. Then, green light-emitting diodes, blue and green diode lasers were successively developed.
Light Emitting Diode Performance Improvements The luminous efficacy of a light emitting diode is generally referred to as the device's external quantum efficiency. It is the product of the device's internal quantum efficiency and the device's extraction efficiency. The so-called internal quantum efficiency of the device is actually the electro-optical conversion efficiency of the device itself, which is mainly related to the characteristics of the device itself such as the energy band, defects, impurities of the device material, and the epitaxial composition and structure of the device. The removal efficiency of the device refers to the number of photons generated inside the device, which is actually measured outside the device after absorption, refraction, and reflection by the device itself. Therefore, factors related to the removal efficiency include the absorption of the device material itself, the geometry of the device, the refractive index difference between the device and the package material, and the scattering characteristics of the device structure. The product of the above two kinds of performance is the luminous effect of the entire device, that is, the external quantum efficiency of the device. The development of early devices focused on improving its internal quantum efficiency. The main method is to improve the quality of epitaxial crystals and change the structure of epitaxial crystals so that the electrical energy is not easily converted into thermal energy, which indirectly improves the luminous efficacy of LEDs, and can be obtained about 70% or so. The theoretical internal quantum efficiency. However, this internal quantum efficiency is almost close to the theoretical limit. Under such circumstances, it is impossible to increase the total light quantity of the device by merely increasing the internal quantum efficiency of the device. That is, the external quantum efficiency reaches two or three times the current quantum efficiency. Enhancing the removal efficiency of the device has become an important issue. The current method for improving device removal efficiency can be mainly divided into the following directions:
Changes in Grain Shape - TIP Structure The fabrication of conventional LED die is a standard rectangular appearance. Because the difference between the refractive index of the general semiconductor material and the epoxy resin is large, the critical angle of total reflection at the interface is small, and the four cross sections of the rectangles are parallel to each other, and the probability of the photon leaving the semiconductor at the interface becomes smaller, so that the photon can only be The internal total reflection until absorbed is exhausted, causing the light to turn into heat, resulting in poorer luminescence. Therefore, changing the shape of the LED is an effective way to improve luminous efficacy. The TIP (Truncated-Inverted-Pyramid) grain structure developed by HP company, the four cross-sections will no longer be parallel to each other, and the light can be effectively extracted, and the external quantum efficiency is substantially increased to 55%. With a performance of up to 100 lm/W, it is the first LED to reach 100 lm/W (Figure 3).
However, HP's TIPLEDs are used only on quaternary red light-emitting diodes that are easy to process. This is particularly difficult for GaN-based light-emitting diodes that use extremely rigid sapphire substrates. In early 2001, Cree made the same structural concept (Figure 4) sandwiching its substrate as an advantage of SiC. It also succeeded in making GaN/SiC light-emitting diodes the same as beveled LEDs, and greatly increased the external quantum efficiency to 32%; however, SiC substrates are much more expensive than Sapphire, so there is no further progress in this technology.
Surface roughness
By roughening the internal and external geometries of the device, the total reflection of the light inside the device is destroyed, and the device's removal efficiency is improved. This method was first proposed by Nichia Chemical. The roughening method basically forms regular irregular shapes on the geometry of the device. The structure of this regular distribution is also divided into two forms depending on the location of the device. One is to provide concave and convex shapes in the device, and the other is to form a regular concave-convex shape over the device and to provide a reflective layer on the back of the device. Since the concave-convex shape can be provided at the interface of the GaN-based compound semiconductor layer using a conventional process, the above-described first method has high practicality. At present, if an ultraviolet device with a wavelength of 405 nm is used, 43% of the external quantum efficiency can be obtained, and the removal efficiency is 60%, which is the highest external quantum efficiency and removal efficiency in the world.
Wafer bonding technology
Since the light generated by the light emitting diode undergoes multiple total reflections, most of the light is absorbed by the semiconductor material itself and the packaging material. Therefore, if GaAs that absorbs light is used as the substrate of the AlGaInPLED, the absorption loss inside the light emitting diode will become larger and the light extraction efficiency of the device will be reduced. In order to reduce the absorption of light emitted from the LED by the substrate, HP first proposed a transparent substrate paste technology. The so-called transparent substrate sticking technology mainly applies pressure on the light emitting diode die under a high temperature environment first, and sticks the transparent GaP substrate, and then removes the GaAs, so that the light extraction rate can be doubled.
The above-mentioned chip sticking technology is mainly applied to quaternary LED devices at present, but recently this technology has also begun to be applied to GaN LEDs. OsramOpto Semiconductors also released a new research result in February 2003 - ThinGaN, can increase the efficiency of blue LED light extraction to 75%, increased by 3 times than the conventional.
Flipchip
For a GaN-based material using a sapphire substrate, since its P- and N-pole electrodes must be on the same side of the device, if a conventional packaging method is used, an upper light-emitting surface that occupies most of the device's light-emission angle is used. A considerable amount of light will be lost due to the blocking of the electrodes. The so-called flipchip structure is to reverse the conventional device and make a reflective layer with a higher reflectivity above the p-type electrode, so that the light originally emitted from above the device is derived from the other light emitting angles of the device, and the sapphire substrate side Edge to take light (Figure 5). This method can reduce the light loss at the electrode side, and can have a light output that is approximately twice that of the conventional packaging method. On the other hand, because the flip-chip structure can directly contact the heat dissipation structure in the package structure directly through the electrodes or bumps, the heat dissipation effect of the device is greatly improved, and the light quantity of the device is further increased.
Solid-state lighting - the light-emitting principle of white light-emitting diodes While the luminous efficacy of colored LEDs has begun to increase substantially, the possibility of applying high-brightness LEDs to lighting is also increasing. The consideration of this application is how to develop a white light emitting diode.
At present, there are mainly three methods for using white light with light emitting diodes, which are described as follows:
Monocrystalline blue LED and yellow fluorescent powder Nichia after the successful development of blue light semiconductors, followed by the development of the product is a white light-emitting diodes. Actually, the white light emitting diode of Nichia does not directly emit white light from the semiconductor material itself. Instead, the yellow light YAG phosphor coated on top of it is excited by the blue light emitting diode. The yellow light generated after excitation of the fluorescent powder and the original light are used. The excited blue light is complementary to generate white light. At present, Nissan's commercial products use 460nm InGaN blue light semiconductor to excite YAG fluorescent powder to generate 555nm yellow light, and it has been fully commercialized. It is similar to several other companies that are developing high-brightness LEDs, such as LumiledsLighting. Cree and Toyoda Gosei continue to compete in the LED market. With the continuous improvement of the luminous efficiency of blue crystal grains and the gradual maturation of YAG phosphor powder synthesis technology, white light-emitting diodes encapsulated with blue crystals and yellow fluorescent powder are currently mature white light-emitting diode technologies.
Single crystal type UVLED+RGB fluorescent powder Although the white light emitting diode packaging technology using blue crystal grains combined with yellow YAG fluorescent powder is a relatively mature technology at present, the white light emitting diodes packaged by using this method have several serious problems. Can't solve it late.
The first is the problem of uniformity because the blue crystal grains that excite the yellow phosphor actually participate in the white color matching. Therefore, the shift in the emission wavelength of the blue crystal grains, the change in the intensity, and the change in the coating thickness of the phosphor powder all affect the uniformity of the white light. degree. The most commonly seen example is that the central part of the white light-emitting diode encapsulated in this way looks blue (or white), while the adjacent area looks yellow (the phosphor powder is thicker). The color of each white light emitting diode is even more different.
On the other hand, the Nichia company that developed this technology owns most of the patents related to the blue-crystal grain process technology and the yellow-light YAG phosphor-related white light-emitting diodes, while the Nichia company adopts an oligopolistic attitude toward patents, and therefore The manufacturers of blue light-emitting diodes using blue light crystals and yellow fluorescent powder are all suffering. The use of blue light crystals coupled with yellow light fluorescent white light emitting diode technology, more white color temperature is high, low color rendering and other issues. Therefore, the development of a technology that is more effective and has no patent issues is a major issue for current LED manufacturers.
UVLED coupled with three color (R, G, B) fluorescent powder provides another development direction. The method mainly uses the UV LEDs that do not actually participate in the distribution of white light to excite the red, green, and blue three-color phosphor powders, and the three-color light emitted by the three-color phosphor powders becomes white light. This method is because the UVLED is not actually involved in the color matching of white light, so the fluctuation of the wavelength and intensity of the UVLED is not particularly sensitive to the white light that is being dispensed. The color temperature and color rendering white light can be modulated through the selection and ratio of fluorescent powders. In terms of patents, the research and development related to the use of UVLED+RGB phosphors still have considerable room for development. However, although such technologies have various advantages, they still have considerable technical difficulties. These difficulties include the selection of the wavelength of the ultraviolet light of the fluorescent powder (excitation wavelength of the best conversion efficiency of the fluorescent powder), the difficulty of UV LED production, and the resistance to UV. The development of packaging materials, etc., has to be addressed by all R&D units.
Polycrystalline RGBLED
The crystals emitting three colors of red, blue, and green are packaged together, and white light-emitting diodes can be made by directly arranging white light in three colors of red, green, and blue. The use of three-color crystals directly encapsulated into white light diodes is the first method used to make white light. Its advantage is that it does not require conversion of phosphor powder. It is directly formed into white light through three-color crystal grains, except that fluorescence can be avoided. In addition to the better luminous performance due to the powder conversion loss, it is also possible to control the light intensity of the three-color light-emitting diodes to achieve a full-color color change effect (color changeable temperature), and can be obtained through the selection of the crystal wavelength and intensity. Good color rendering. However, its disadvantage is that the light mixing is difficult. The user can easily observe a variety of different colors in front of the light source, and see colored shadows behind each shield. In addition, because the three dies used are all heat sources, the heat dissipation problem is three times that of other types of packages, thereby increasing the difficulty of its use. Currently, the use of a polycrystalline RGB LED package type white light emitting diode can achieve a performance of about 25-30 lm/W. It is mainly used in outdoor display panels, outdoor landscape lights, and color washable wall lamps where the heat dissipation problem is less serious. On the other hand, if the design can be controlled by an electronic circuit, the use of a polycrystalline RGB LED package type of light emitting diode is likely to be one of the main sources of light replacing the backlight in LCD backlight modules currently using CCFL.
Cooling is the main research topic of white light led lighting. Although the luminous efficiency of white light-emitting diodes is gradually increasing, the possibility of applying white light-emitting diodes to lighting is also increasing, but it is clear that a single white light-emitting diode drives The power supply is low, so with the current package type it is unlikely that a single white light-emitting diode will be required to achieve the lumens required for illumination. In response to this problem, the main solutions currently can be roughly divided into two types. One is to use a plurality of light emitting diodes to form a light source module than conventionally, and each single light emitting diode needs a driving power source and a common one. The same (about 20-30mA) is used; the other method is currently used by several high-brightness LED manufacturers, that is, using the so-called large-grain process where the conventional die size is no longer used. (0.3 mm), but the grain process size is larger (0.6 mu -1 mm), and such a light emitting device is driven with a high driving current (typically 150-350 mA, currently up to 500 mA or more). However, no matter what method is used, it will be necessary to handle extremely high heat in a very small LED package. If the device cannot disperse these high heat, besides the various packaging materials, the reliability of the product will be different due to the difference in the expansion coefficient between them. In addition to the problem of degree, the luminous efficiency of the crystal grains also decreases significantly with the increase of the temperature, and the lifetime of the crystal grains is significantly shortened. Therefore, how to dissipate the high heat in the device has become an important issue in the current light emitting diode packaging technology.
For a light emitting diode, the most important one is the output light flux and light shape, so one end of the light emitting diode must not be blocked, but need to be coated with a highly transparent epoxy resin material. However, the current epoxy resins are almost non-thermally conductive materials. Therefore, for the current LED packaging technology, the main heat dissipation is to use a leadframe under the light emitting diode die to disperse the device. The heat. As far as the current trend is concerned, the selection of metal leg materials is mainly based on materials with high thermal conductivity, such as aluminum, copper, and even ceramic materials. However, the thermal expansion coefficient between these materials and grains is very different, if they are directly contacted. It is likely that the problem of reliability arises from the stress generated between materials when the temperature rises. Therefore, an intermediate material having both the coefficient of conduct and the coefficient of expansion is generally used as the pitch between the materials. Using the above concept, Matsushita Electric Co., Ltd. made several light-emitting diodes on multi-layered substrate modules made of metal materials and metal-based composite materials in 2003 to form a light source module, utilizing the high thermal conductivity of the light source substrate to make light sources The output remains stable over long periods of use (see Figure 6).
Also using the idea of a high heat-dissipating substrate, Lumileds applied it to large-area-grained products (see Figure 7). The material used in the Lumileds substrate is a copper material with a high conductivity coefficient, which is then connected to a special metal circuit board, taking into account the conduction of the circuit and increasing the effect of heat transfer.
In addition to Lumileds, products such as Osram Opto Semiconductors and Nichia Chemical have introduced large-grain products with a size of more than 1 W (Figures 8 and 9). From these high-brightness light-emitting diode manufacturers have launched large-grain, high-power products, it seems that large-grain-related processes, packaging technology seems to have gradually become the mainstream of high-brightness light-emitting diodes. However, large-grain-related processes and packaging technologies are not just increasing the die area. Related processes and packaging technologies still have considerable barriers for conventional LED manufacturers, but if you want to push LEDs to high-brightness lighting The research and development of related fields and technologies are still necessary.
Turning technology into mass production As the light-emitting efficiency of light-emitting diodes has gradually increased in recent years, the possibility of using light-emitting diodes as light-emitting sources is also increasing. However, when people only consider improving the light-emitting efficiency of light-emitting diodes, how to make full use of the characteristics of the light-emitting diodes and solve the difficulties that may be encountered when applying them to lighting are already the current goals of major lighting manufacturers. Current difficulties include heat dissipation and the use of special light emitting diodes for light emitting diodes.
In terms of heat dissipation, although the light emitting diode is known as a cold light source, there is still room for improvement in its electro-optical performance. That is to say, there is still a considerable amount of electrical energy that is not converted into light and causes excess heat energy. These heat energy is concentrated in the crystal grains. The size will cause severe heat dissipation problems. Therefore, good heat dissipation design and development of heat-dissipation materials are currently the focus.
In terms of light emitting diode light emitting diodes, light emitting diodes have completely different light emitting characteristics from conventional light sources. In addition to the extremely small size of the die itself, various types of light emitting diode packages also cause completely different types of light emitting diodes. Because of its light-emitting shape, a design relative to a light-emitting diode lighting application will no longer be able to simply put a condenser lens or a mirror on the light source, but rather must undergo a more careful optical design. In these parts of research and development, companies and research and development units have different directions, but in addition to developing technologies, how can these technologies be mass-produced and reduce the cost of these solid-state light sources? In the next few years, solid-state light sources can become lighting sources. The key to mainstream
